• The Legal Accountability of AI-Generated Deepfakes in Election Misinformation

    How Deepfakes Are Created

    Generative AI models enable the creation of highly realistic fake media. Most deepfakes today are produced by training deep neural networks on real images, video or audio of a target person. The two predominant AI architectures are generative adversarial networksand autoencoders. A GAN consists of a generator network that produces synthetic images and a discriminator network that tries to distinguish fakes from real data. Through iterative training, the generator learns to produce outputs that increasingly fool the discriminator¹. Autoencoder-based tools similarly learn to encode a target face and then decode it onto a source video. In practice, deepfake creators use accessible software: open-source tools like DeepFaceLab and FaceSwap dominate video face-swapping². Voice-cloning toolscan mimic a person’s speech from minutes of audio. Commercial platforms like Synthesia allow text-to-video avatars, which have already been misused in disinformation campaigns³. Even mobile appslet users do basic face swaps in minutes⁴. In short, advances in GANs and related models make deepfakes cheaper and easier to generate than ever.

    Diagram of a generative adversarial network: A generator network creates fake images from random input and a discriminator network distinguishes fakes from real examples. Over time the generator improves until its outputs “fool” the discriminator⁵

    During creation, a deepfake algorithm is typically trained on a large dataset of real images or audio from the target. The more varied and high-quality the training data, the more realistic the deepfake. The output often then undergoes post-processingto enhance believability¹. Technical defenses focus on two fronts: detection and authentication. Detection uses AI models to spot inconsistenciesthat betray a synthetic origin⁵. Authentication embeds markers before dissemination – for example, invisible watermarks or cryptographically signed metadata indicating authenticity⁶. The EU AI Act will soon mandate that major AI content providers embed machine-readable “watermark” signals in synthetic media⁷. However, as GAO notes, detection is an arms race – even a marked deepfake can sometimes evade notice – and labels alone don’t stop false narratives from spreading⁸⁹.

    Deepfakes in Recent Elections: Examples

    Deepfakes and AI-generated imagery already have made headlines in election cycles around the world. In the 2024 U.S. primary season, a digitally-altered audio robocall mimicked President Biden’s voice urging Democrats not to vote in the New Hampshire primary. The callerwas later fined million by the FCC and indicted under existing telemarketing laws¹⁰¹¹.Also in 2024, former President Trump posted on social media a collage implying that pop singer Taylor Swift endorsed his campaign, using AI-generated images of Swift in “Swifties for Trump” shirts¹². The posts sparked media uproar, though analysts noted the same effect could have been achieved without AI¹². Similarly, Elon Musk’s X platform carried AI-generated clips, including a parody “Ad” depicting Vice-President Harris’s voice via an AI clone¹³.

    Beyond the U.S., deepfake-like content has appeared globally. In Indonesia’s 2024 presidential election, a video surfaced on social media in which a convincingly generated image of the late President Suharto appeared to endorse the candidate of the Golkar Party. Days later, the endorsed candidatewon the presidency¹⁴. In Bangladesh, a viral deepfake video superimposed the face of opposition leader Rumeen Farhana onto a bikini-clad body – an incendiary fabrication designed to discredit her in the conservative Muslim-majority society¹⁵. Moldova’s pro-Western President Maia Sandu has been repeatedly targeted by AI-driven disinformation; one deepfake video falsely showed her resigning and endorsing a Russian-friendly party, apparently to sow distrust in the electoral process¹⁶. Even in Taiwan, a TikTok clip circulated that synthetically portrayed a U.S. politician making foreign-policy statements – stoking confusion ahead of Taiwanese elections¹⁷. In Slovakia’s recent campaign, AI-generated audio mimicking the liberal party leader suggested he plotted vote-rigging and beer-price hikes – instantly spreading on social media just days before the election¹⁸. These examples show that deepfakes have touched diverse polities, often aiming to undermine candidates or confuse voters¹⁵¹⁸.

    Notably, many of the most viral “deepfakes” in 2024 were actually circulated as obvious memes or claims, rather than subtle deceptions. Experts observed that outright undetectable AI deepfakes were relatively rare; more common were AI-generated memes plainly shared by partisans, or cheaply doctored “cheapfakes” made with basic editing tools¹³¹⁹. For instance, social media was awash with memes of Kamala Harris in Soviet garb or of Black Americans holding Trump signs¹³, but these were typically used satirically, not meant to be secretly believed. Nonetheless, even unsophisticated fakes can sway opinion: a U.S. study found that false presidential adsdid change voter attitudes in swing states. In sum, deepfakes are a real and growing phenomenon in election campaigns²⁰²¹ worldwide – a trend taken seriously by voters and regulators alike.

    U.S. Legal Framework and Accountability

    In the U.S., deepfake creators and distributors of election misinformation face a patchwork of tools, but no single comprehensive federal “deepfake law.” Existing laws relevant to disinformation include statutes against impersonating government officials, electioneering, and targeted statutes like criminal electioneering communications. In some cases ordinary laws have been stretched: the NH robocall used the Telephone Consumer Protection Act and mail/telemarketing fraud provisions, resulting in the M fine and a criminal charge. Similarly, voice impostors can potentially violate laws against “false advertising” or “unlawful corporate communications.” However, these laws were enacted before AI, and litigators have warned they often do not fit neatly. For example, deceptive deepfake claims not tied to a specific victim do not easily fit into defamation or privacy torts. Voter intimidation lawsalso leave a gap for non-threatening falsehoods about voting logistics or endorsements.

    Recognizing these gaps, some courts and agencies are invoking other theories. The U.S. Department of Justice has recently charged individuals under broad fraud statutes, and state attorneys general have considered deepfake misinformation as interference with voting rights. Notably, the Federal Election Commissionis preparing to enforce new rules: in April 2024 it issued an advisory opinion limiting “non-candidate electioneering communications” that use falsified media, effectively requiring that political ads use only real images of the candidate. If finalized, that would make it unlawful for campaigns to pay for ads depicting a candidate saying things they never did. Similarly, the Federal Trade Commissionand Department of Justicehave signaled that purely commercial deepfakes could violate consumer protection or election laws.

    U.S. Legislation and Proposals

    Federal lawmakers have proposed new statutes. The DEEPFAKES Accountability Actwould, among other things, impose a disclosure requirement: political ads featuring a manipulated media likeness would need clear disclaimers identifying the content as synthetic. It also increases penalties for producing false election videos or audio intended to influence the vote. While not yet enacted, supporters argue it would provide a uniform rule for all federal and state campaigns. The Brennan Center supports transparency requirements over outright bans, suggesting laws should narrowly target deceptive deepfakes in paid ads or certain categorieswhile carving out parody and news coverage.

    At the state level, over 20 states have passed deepfake laws specifically for elections. For example, Florida and California forbid distributing falsified audio/visual media of candidates with intent to deceive voters. Some statesdefine “deepfake” in statutes and allow candidates to sue or revoke candidacies of violators. These measures have had mixed success: courts have struck down overly broad provisions that acted as prior restraints. Critically, these state laws raise First Amendment issues: political speech is highly protected, so any restriction must be tightly tailored. Already, Texas and Virginia statutes are under legal review, and Elon Musk’s company has sued under California’s lawas unconstitutional. In practice, most lawsuits have so far centered on defamation or intellectual property, rather than election-focused statutes.

    Policy Recommendations: Balancing Integrity and Speech

    Given the rapidly evolving technology, experts recommend a multi-pronged approach. Most stress transparency and disclosure as core principles. For example, the Brennan Center urges requiring any political communication that uses AI-synthesized images or voice to include a clear label. This could be a digital watermark or a visible disclaimer. Transparency has two advantages: it forces campaigns and platforms to “own” the use of AI, and it alerts audiences to treat the content with skepticism.

    Outright bans on all deepfakes would likely violate free speech, but targeted bans on specific harmsmay be defensible. Indeed, Florida already penalizes misuse of recordings in voter suppression. Another recommendation is limited liability: tying penalties to demonstrable intent to mislead, not to the mere act of content creation. Both U.S. federal proposals and EU law generally condition fines on the “appearance of fraud” or deception.

    Technical solutions can complement laws. Watermarking original mediacould deter the reuse of authentic images in doctored fakes. Open tools for deepfake detection – some supported by government research grants – should be deployed by fact-checkers and social platforms. Making detection datasets publicly availablehelps improve AI models to spot fakes. International cooperation is also urged: cross-border agreements on information-sharing could help trace and halt disinformation campaigns. The G7 and APEC have all recently committed to fighting election interference via AI, which may lead to joint norms or rapid response teams.

    Ultimately, many analysts believe the strongest “cure” is a well-informed public: education campaigns to teach voters to question sensational media, and a robust independent press to debunk falsehoods swiftly. While the law can penalize the worst offenders, awareness and resilience in the electorate are crucial buffers against influence operations. As Georgia Tech’s Sean Parker quipped in 2019, “the real question is not if deepfakes will influence elections, but who will be empowered by the first effective one.” Thus policies should aim to deter malicious use without unduly chilling innovation or satire.

    References:

    /.

    /.

    .

    .

    .

    .

    .

    .

    .

    /.

    .

    .

    /.

    /.

    .

    The post The Legal Accountability of AI-Generated Deepfakes in Election Misinformation appeared first on MarkTechPost.
    #legal #accountability #aigenerated #deepfakes #election
    The Legal Accountability of AI-Generated Deepfakes in Election Misinformation
    How Deepfakes Are Created Generative AI models enable the creation of highly realistic fake media. Most deepfakes today are produced by training deep neural networks on real images, video or audio of a target person. The two predominant AI architectures are generative adversarial networksand autoencoders. A GAN consists of a generator network that produces synthetic images and a discriminator network that tries to distinguish fakes from real data. Through iterative training, the generator learns to produce outputs that increasingly fool the discriminator¹. Autoencoder-based tools similarly learn to encode a target face and then decode it onto a source video. In practice, deepfake creators use accessible software: open-source tools like DeepFaceLab and FaceSwap dominate video face-swapping². Voice-cloning toolscan mimic a person’s speech from minutes of audio. Commercial platforms like Synthesia allow text-to-video avatars, which have already been misused in disinformation campaigns³. Even mobile appslet users do basic face swaps in minutes⁴. In short, advances in GANs and related models make deepfakes cheaper and easier to generate than ever. Diagram of a generative adversarial network: A generator network creates fake images from random input and a discriminator network distinguishes fakes from real examples. Over time the generator improves until its outputs “fool” the discriminator⁵ During creation, a deepfake algorithm is typically trained on a large dataset of real images or audio from the target. The more varied and high-quality the training data, the more realistic the deepfake. The output often then undergoes post-processingto enhance believability¹. Technical defenses focus on two fronts: detection and authentication. Detection uses AI models to spot inconsistenciesthat betray a synthetic origin⁵. Authentication embeds markers before dissemination – for example, invisible watermarks or cryptographically signed metadata indicating authenticity⁶. The EU AI Act will soon mandate that major AI content providers embed machine-readable “watermark” signals in synthetic media⁷. However, as GAO notes, detection is an arms race – even a marked deepfake can sometimes evade notice – and labels alone don’t stop false narratives from spreading⁸⁹. Deepfakes in Recent Elections: Examples Deepfakes and AI-generated imagery already have made headlines in election cycles around the world. In the 2024 U.S. primary season, a digitally-altered audio robocall mimicked President Biden’s voice urging Democrats not to vote in the New Hampshire primary. The callerwas later fined million by the FCC and indicted under existing telemarketing laws¹⁰¹¹.Also in 2024, former President Trump posted on social media a collage implying that pop singer Taylor Swift endorsed his campaign, using AI-generated images of Swift in “Swifties for Trump” shirts¹². The posts sparked media uproar, though analysts noted the same effect could have been achieved without AI¹². Similarly, Elon Musk’s X platform carried AI-generated clips, including a parody “Ad” depicting Vice-President Harris’s voice via an AI clone¹³. Beyond the U.S., deepfake-like content has appeared globally. In Indonesia’s 2024 presidential election, a video surfaced on social media in which a convincingly generated image of the late President Suharto appeared to endorse the candidate of the Golkar Party. Days later, the endorsed candidatewon the presidency¹⁴. In Bangladesh, a viral deepfake video superimposed the face of opposition leader Rumeen Farhana onto a bikini-clad body – an incendiary fabrication designed to discredit her in the conservative Muslim-majority society¹⁵. Moldova’s pro-Western President Maia Sandu has been repeatedly targeted by AI-driven disinformation; one deepfake video falsely showed her resigning and endorsing a Russian-friendly party, apparently to sow distrust in the electoral process¹⁶. Even in Taiwan, a TikTok clip circulated that synthetically portrayed a U.S. politician making foreign-policy statements – stoking confusion ahead of Taiwanese elections¹⁷. In Slovakia’s recent campaign, AI-generated audio mimicking the liberal party leader suggested he plotted vote-rigging and beer-price hikes – instantly spreading on social media just days before the election¹⁸. These examples show that deepfakes have touched diverse polities, often aiming to undermine candidates or confuse voters¹⁵¹⁸. Notably, many of the most viral “deepfakes” in 2024 were actually circulated as obvious memes or claims, rather than subtle deceptions. Experts observed that outright undetectable AI deepfakes were relatively rare; more common were AI-generated memes plainly shared by partisans, or cheaply doctored “cheapfakes” made with basic editing tools¹³¹⁹. For instance, social media was awash with memes of Kamala Harris in Soviet garb or of Black Americans holding Trump signs¹³, but these were typically used satirically, not meant to be secretly believed. Nonetheless, even unsophisticated fakes can sway opinion: a U.S. study found that false presidential adsdid change voter attitudes in swing states. In sum, deepfakes are a real and growing phenomenon in election campaigns²⁰²¹ worldwide – a trend taken seriously by voters and regulators alike. U.S. Legal Framework and Accountability In the U.S., deepfake creators and distributors of election misinformation face a patchwork of tools, but no single comprehensive federal “deepfake law.” Existing laws relevant to disinformation include statutes against impersonating government officials, electioneering, and targeted statutes like criminal electioneering communications. In some cases ordinary laws have been stretched: the NH robocall used the Telephone Consumer Protection Act and mail/telemarketing fraud provisions, resulting in the M fine and a criminal charge. Similarly, voice impostors can potentially violate laws against “false advertising” or “unlawful corporate communications.” However, these laws were enacted before AI, and litigators have warned they often do not fit neatly. For example, deceptive deepfake claims not tied to a specific victim do not easily fit into defamation or privacy torts. Voter intimidation lawsalso leave a gap for non-threatening falsehoods about voting logistics or endorsements. Recognizing these gaps, some courts and agencies are invoking other theories. The U.S. Department of Justice has recently charged individuals under broad fraud statutes, and state attorneys general have considered deepfake misinformation as interference with voting rights. Notably, the Federal Election Commissionis preparing to enforce new rules: in April 2024 it issued an advisory opinion limiting “non-candidate electioneering communications” that use falsified media, effectively requiring that political ads use only real images of the candidate. If finalized, that would make it unlawful for campaigns to pay for ads depicting a candidate saying things they never did. Similarly, the Federal Trade Commissionand Department of Justicehave signaled that purely commercial deepfakes could violate consumer protection or election laws. U.S. Legislation and Proposals Federal lawmakers have proposed new statutes. The DEEPFAKES Accountability Actwould, among other things, impose a disclosure requirement: political ads featuring a manipulated media likeness would need clear disclaimers identifying the content as synthetic. It also increases penalties for producing false election videos or audio intended to influence the vote. While not yet enacted, supporters argue it would provide a uniform rule for all federal and state campaigns. The Brennan Center supports transparency requirements over outright bans, suggesting laws should narrowly target deceptive deepfakes in paid ads or certain categorieswhile carving out parody and news coverage. At the state level, over 20 states have passed deepfake laws specifically for elections. For example, Florida and California forbid distributing falsified audio/visual media of candidates with intent to deceive voters. Some statesdefine “deepfake” in statutes and allow candidates to sue or revoke candidacies of violators. These measures have had mixed success: courts have struck down overly broad provisions that acted as prior restraints. Critically, these state laws raise First Amendment issues: political speech is highly protected, so any restriction must be tightly tailored. Already, Texas and Virginia statutes are under legal review, and Elon Musk’s company has sued under California’s lawas unconstitutional. In practice, most lawsuits have so far centered on defamation or intellectual property, rather than election-focused statutes. Policy Recommendations: Balancing Integrity and Speech Given the rapidly evolving technology, experts recommend a multi-pronged approach. Most stress transparency and disclosure as core principles. For example, the Brennan Center urges requiring any political communication that uses AI-synthesized images or voice to include a clear label. This could be a digital watermark or a visible disclaimer. Transparency has two advantages: it forces campaigns and platforms to “own” the use of AI, and it alerts audiences to treat the content with skepticism. Outright bans on all deepfakes would likely violate free speech, but targeted bans on specific harmsmay be defensible. Indeed, Florida already penalizes misuse of recordings in voter suppression. Another recommendation is limited liability: tying penalties to demonstrable intent to mislead, not to the mere act of content creation. Both U.S. federal proposals and EU law generally condition fines on the “appearance of fraud” or deception. Technical solutions can complement laws. Watermarking original mediacould deter the reuse of authentic images in doctored fakes. Open tools for deepfake detection – some supported by government research grants – should be deployed by fact-checkers and social platforms. Making detection datasets publicly availablehelps improve AI models to spot fakes. International cooperation is also urged: cross-border agreements on information-sharing could help trace and halt disinformation campaigns. The G7 and APEC have all recently committed to fighting election interference via AI, which may lead to joint norms or rapid response teams. Ultimately, many analysts believe the strongest “cure” is a well-informed public: education campaigns to teach voters to question sensational media, and a robust independent press to debunk falsehoods swiftly. While the law can penalize the worst offenders, awareness and resilience in the electorate are crucial buffers against influence operations. As Georgia Tech’s Sean Parker quipped in 2019, “the real question is not if deepfakes will influence elections, but who will be empowered by the first effective one.” Thus policies should aim to deter malicious use without unduly chilling innovation or satire. References: /. /. . . . . . . . /. . . /. /. . The post The Legal Accountability of AI-Generated Deepfakes in Election Misinformation appeared first on MarkTechPost. #legal #accountability #aigenerated #deepfakes #election
    WWW.MARKTECHPOST.COM
    The Legal Accountability of AI-Generated Deepfakes in Election Misinformation
    How Deepfakes Are Created Generative AI models enable the creation of highly realistic fake media. Most deepfakes today are produced by training deep neural networks on real images, video or audio of a target person. The two predominant AI architectures are generative adversarial networks (GANs) and autoencoders. A GAN consists of a generator network that produces synthetic images and a discriminator network that tries to distinguish fakes from real data. Through iterative training, the generator learns to produce outputs that increasingly fool the discriminator¹. Autoencoder-based tools similarly learn to encode a target face and then decode it onto a source video. In practice, deepfake creators use accessible software: open-source tools like DeepFaceLab and FaceSwap dominate video face-swapping (one estimate suggests DeepFaceLab was used for over 95% of known deepfake videos)². Voice-cloning tools (often built on similar AI principles) can mimic a person’s speech from minutes of audio. Commercial platforms like Synthesia allow text-to-video avatars (turning typed scripts into lifelike “spokespeople”), which have already been misused in disinformation campaigns³. Even mobile apps (e.g. FaceApp, Zao) let users do basic face swaps in minutes⁴. In short, advances in GANs and related models make deepfakes cheaper and easier to generate than ever. Diagram of a generative adversarial network (GAN): A generator network creates fake images from random input and a discriminator network distinguishes fakes from real examples. Over time the generator improves until its outputs “fool” the discriminator⁵ During creation, a deepfake algorithm is typically trained on a large dataset of real images or audio from the target. The more varied and high-quality the training data, the more realistic the deepfake. The output often then undergoes post-processing (color adjustments, lip-syncing refinements) to enhance believability¹. Technical defenses focus on two fronts: detection and authentication. Detection uses AI models to spot inconsistencies (blinking irregularities, audio artifacts or metadata mismatches) that betray a synthetic origin⁵. Authentication embeds markers before dissemination – for example, invisible watermarks or cryptographically signed metadata indicating authenticity⁶. The EU AI Act will soon mandate that major AI content providers embed machine-readable “watermark” signals in synthetic media⁷. However, as GAO notes, detection is an arms race – even a marked deepfake can sometimes evade notice – and labels alone don’t stop false narratives from spreading⁸⁹. Deepfakes in Recent Elections: Examples Deepfakes and AI-generated imagery already have made headlines in election cycles around the world. In the 2024 U.S. primary season, a digitally-altered audio robocall mimicked President Biden’s voice urging Democrats not to vote in the New Hampshire primary. The caller (“Susan Anderson”) was later fined $6 million by the FCC and indicted under existing telemarketing laws¹⁰¹¹. (Importantly, FCC rules on robocalls applied regardless of AI: the perpetrator could have used a voice actor or recording instead.) Also in 2024, former President Trump posted on social media a collage implying that pop singer Taylor Swift endorsed his campaign, using AI-generated images of Swift in “Swifties for Trump” shirts¹². The posts sparked media uproar, though analysts noted the same effect could have been achieved without AI (e.g., by photoshopping text on real images)¹². Similarly, Elon Musk’s X platform carried AI-generated clips, including a parody “Ad” depicting Vice-President Harris’s voice via an AI clone¹³. Beyond the U.S., deepfake-like content has appeared globally. In Indonesia’s 2024 presidential election, a video surfaced on social media in which a convincingly generated image of the late President Suharto appeared to endorse the candidate of the Golkar Party. Days later, the endorsed candidate (who is Suharto’s son-in-law) won the presidency¹⁴. In Bangladesh, a viral deepfake video superimposed the face of opposition leader Rumeen Farhana onto a bikini-clad body – an incendiary fabrication designed to discredit her in the conservative Muslim-majority society¹⁵. Moldova’s pro-Western President Maia Sandu has been repeatedly targeted by AI-driven disinformation; one deepfake video falsely showed her resigning and endorsing a Russian-friendly party, apparently to sow distrust in the electoral process¹⁶. Even in Taiwan (amidst tensions with China), a TikTok clip circulated that synthetically portrayed a U.S. politician making foreign-policy statements – stoking confusion ahead of Taiwanese elections¹⁷. In Slovakia’s recent campaign, AI-generated audio mimicking the liberal party leader suggested he plotted vote-rigging and beer-price hikes – instantly spreading on social media just days before the election¹⁸. These examples show that deepfakes have touched diverse polities (from Bangladesh and Indonesia to Moldova, Slovakia, India and beyond), often aiming to undermine candidates or confuse voters¹⁵¹⁸. Notably, many of the most viral “deepfakes” in 2024 were actually circulated as obvious memes or claims, rather than subtle deceptions. Experts observed that outright undetectable AI deepfakes were relatively rare; more common were AI-generated memes plainly shared by partisans, or cheaply doctored “cheapfakes” made with basic editing tools¹³¹⁹. For instance, social media was awash with memes of Kamala Harris in Soviet garb or of Black Americans holding Trump signs¹³, but these were typically used satirically, not meant to be secretly believed. Nonetheless, even unsophisticated fakes can sway opinion: a U.S. study found that false presidential ads (not necessarily AI-made) did change voter attitudes in swing states. In sum, deepfakes are a real and growing phenomenon in election campaigns²⁰²¹ worldwide – a trend taken seriously by voters and regulators alike. U.S. Legal Framework and Accountability In the U.S., deepfake creators and distributors of election misinformation face a patchwork of tools, but no single comprehensive federal “deepfake law.” Existing laws relevant to disinformation include statutes against impersonating government officials, electioneering (such as the Bipartisan Campaign Reform Act, which requires disclaimers on political ads), and targeted statutes like criminal electioneering communications. In some cases ordinary laws have been stretched: the NH robocall used the Telephone Consumer Protection Act and mail/telemarketing fraud provisions, resulting in the $6M fine and a criminal charge. Similarly, voice impostors can potentially violate laws against “false advertising” or “unlawful corporate communications.” However, these laws were enacted before AI, and litigators have warned they often do not fit neatly. For example, deceptive deepfake claims not tied to a specific victim do not easily fit into defamation or privacy torts. Voter intimidation laws (prohibiting threats or coercion) also leave a gap for non-threatening falsehoods about voting logistics or endorsements. Recognizing these gaps, some courts and agencies are invoking other theories. The U.S. Department of Justice has recently charged individuals under broad fraud statutes (e.g. for a plot to impersonate an aide to swing votes in 2020), and state attorneys general have considered deepfake misinformation as interference with voting rights. Notably, the Federal Election Commission (FEC) is preparing to enforce new rules: in April 2024 it issued an advisory opinion limiting “non-candidate electioneering communications” that use falsified media, effectively requiring that political ads use only real images of the candidate. If finalized, that would make it unlawful for campaigns to pay for ads depicting a candidate saying things they never did. Similarly, the Federal Trade Commission (FTC) and Department of Justice (DOJ) have signaled that purely commercial deepfakes could violate consumer protection or election laws (for example, liability for mass false impersonation or for foreign-funded electioneering). U.S. Legislation and Proposals Federal lawmakers have proposed new statutes. The DEEPFAKES Accountability Act (H.R.5586 in the 118th Congress) would, among other things, impose a disclosure requirement: political ads featuring a manipulated media likeness would need clear disclaimers identifying the content as synthetic. It also increases penalties for producing false election videos or audio intended to influence the vote. While not yet enacted, supporters argue it would provide a uniform rule for all federal and state campaigns. The Brennan Center supports transparency requirements over outright bans, suggesting laws should narrowly target deceptive deepfakes in paid ads or certain categories (e.g. false claims about time/place/manner of voting) while carving out parody and news coverage. At the state level, over 20 states have passed deepfake laws specifically for elections. For example, Florida and California forbid distributing falsified audio/visual media of candidates with intent to deceive voters (though Florida’s law exempts parody). Some states (like Texas) define “deepfake” in statutes and allow candidates to sue or revoke candidacies of violators. These measures have had mixed success: courts have struck down overly broad provisions that acted as prior restraints (e.g. Minnesota’s 2023 law was challenged for threatening injunctions against anyone “reasonably believed” to violate it). Critically, these state laws raise First Amendment issues: political speech is highly protected, so any restriction must be tightly tailored. Already, Texas and Virginia statutes are under legal review, and Elon Musk’s company has sued under California’s law (which requires platforms to label or block deepfakes) as unconstitutional. In practice, most lawsuits have so far centered on defamation or intellectual property (for instance, a celebrity suing over a botched celebrity-deepfake video), rather than election-focused statutes. Policy Recommendations: Balancing Integrity and Speech Given the rapidly evolving technology, experts recommend a multi-pronged approach. Most stress transparency and disclosure as core principles. For example, the Brennan Center urges requiring any political communication that uses AI-synthesized images or voice to include a clear label. This could be a digital watermark or a visible disclaimer. Transparency has two advantages: it forces campaigns and platforms to “own” the use of AI, and it alerts audiences to treat the content with skepticism. Outright bans on all deepfakes would likely violate free speech, but targeted bans on specific harms (e.g. automated phone calls impersonating voters, or videos claiming false polling information) may be defensible. Indeed, Florida already penalizes misuse of recordings in voter suppression. Another recommendation is limited liability: tying penalties to demonstrable intent to mislead, not to the mere act of content creation. Both U.S. federal proposals and EU law generally condition fines on the “appearance of fraud” or deception. Technical solutions can complement laws. Watermarking original media (as encouraged by the EU AI Act) could deter the reuse of authentic images in doctored fakes. Open tools for deepfake detection – some supported by government research grants – should be deployed by fact-checkers and social platforms. Making detection datasets publicly available (e.g. the MIT OpenDATATEST) helps improve AI models to spot fakes. International cooperation is also urged: cross-border agreements on information-sharing could help trace and halt disinformation campaigns. The G7 and APEC have all recently committed to fighting election interference via AI, which may lead to joint norms or rapid response teams. Ultimately, many analysts believe the strongest “cure” is a well-informed public: education campaigns to teach voters to question sensational media, and a robust independent press to debunk falsehoods swiftly. While the law can penalize the worst offenders, awareness and resilience in the electorate are crucial buffers against influence operations. As Georgia Tech’s Sean Parker quipped in 2019, “the real question is not if deepfakes will influence elections, but who will be empowered by the first effective one.” Thus policies should aim to deter malicious use without unduly chilling innovation or satire. References: https://www.security.org/resources/deepfake-statistics/. https://www.wired.com/story/synthesia-ai-deepfakes-it-control-riparbelli/. https://www.gao.gov/products/gao-24-107292. https://technologyquotient.freshfields.com/post/102jb19/eu-ai-act-unpacked-8-new-rules-on-deepfakes. https://knightcolumbia.org/blog/we-looked-at-78-election-deepfakes-political-misinformation-is-not-an-ai-problem. https://www.npr.org/2024/12/21/nx-s1-5220301/deepfakes-memes-artificial-intelligence-elections. https://apnews.com/article/artificial-intelligence-elections-disinformation-chatgpt-bc283e7426402f0b4baa7df280a4c3fd. https://www.lawfaremedia.org/article/new-and-old-tools-to-tackle-deepfakes-and-election-lies-in-2024. https://www.brennancenter.org/our-work/research-reports/regulating-ai-deepfakes-and-synthetic-media-political-arena. https://firstamendment.mtsu.edu/article/political-deepfakes-and-elections/. https://www.ncsl.org/technology-and-communication/deceptive-audio-or-visual-media-deepfakes-2024-legislation. https://law.unh.edu/sites/default/files/media/2022/06/nagumotu_pp113-157.pdf. https://dfrlab.org/2024/10/02/brazil-election-ai-research/. https://dfrlab.org/2024/11/26/brazil-election-ai-deepfakes/. https://freedomhouse.org/article/eu-digital-services-act-win-transparency. The post The Legal Accountability of AI-Generated Deepfakes in Election Misinformation appeared first on MarkTechPost.
    0 Comentários 0 Compartilhamentos
  • How NPR’s Tiny Desk became the biggest stage in music

    Until last October, Argentinian musical duo Ca7riel & Paco Amoroso were more or less a regional act. Known for their experimental blend of Latin trap, pop, and rap, the pair had a fanbase, but still weren’t cracking more than 3,000 daily streams across services like Spotify, Apple Music, and YouTube. Within a week, they shot up 4,700%—hitting 222,000 daily streams—according to exclusive data firm Luminate, which powers the Billboard charts. Suddenly Ca7riel & Paco Amoroso were global pop stars. 

    What changed? On Oct. 4, the pair were featured in a Tiny Desk Concert, part of NPR’s 17-year-old video series featuring musicians performing stripped-down sets behind an office desk in the cramped Washington, D.C. headquarters of the public broadcaster. 

    In the concert video, the artists play five songs from their debut album Baño Maria, which came out last April. Paco’s raspy voice emerges from underneath a puffy blue trapper hat while Ca7riel sports an over-the-top pout and a vest made of stitched-together heart-shaped plush toys. The pair sing entirely in Spanish, backed by their Argentinian bandmatesand an American horn section. The duo’s performance quickly took off across the internet. Within five days, it had racked up more than 1.5 million views on YouTube, and hit 11 million in little more than a month. It also reverberated across social media: the NPR Music Instagram post garnering nearly 900,000 likes, and TikToks clips garnered hundreds of thousands of views. 

    In a year that featured Tiny Desk performances from buzzy stars like Chappell Roan and Sabrina Carpenter, as well as established acts like Chaka Khan and Nelly Furtado, Ca7riel & Paco Amoroso’s concert was the most-watched of 2024. It currently sits at 36 million views. 

    That virality translated to an influx of bookings for the duo, including a performance at Coachella in April, and upcoming slots at Glastonbury in June, FujiRock Japan in July, and Lollapalooza and Outside Lands in August. Ca7riel & Paco Amoroso’s global tour includes sold-out dates at Mexico’s 20,000-capacity Palacio de los Deportes and Chile’s 14,000-seat Movistar Areana—and was previewed by an appearance on The Tonight Show Starring Jimmy Fallon in April. 

    “Through Tiny Desk, we’ve noticed media approaching us, promoters being very interested in offering their spaces and festivals, and many media outlets opening doors to show us to the world,” says Jonathan Izquierdo, the band’s Spain-based tour manager who began working with the duo shortly after the Tiny Desk Concert debuted. “We’ve managed to sell out summer arena shows in record time and we’re constantly adding new concerts. Promoters are knocking on our doors to get the Tiny Desk effect.”

    Bobby CarterTiny Desk, Big Influence

    The Tiny Desk effect is something Bobby Carter, NPR Tiny Desk host and series producer, has seen firsthand. Carter has been at NPR for 25 years, including the past 11 on the Tiny Desk team. He took the reins when Bob Boilen, the longtime All Songs Considered host who launched Tiny Desk in 2008, retired in 2023. 

    The series—which now has more than 1,200 videos—began as an internet-first way for Boilen to showcase performances from musicians that were more intimate than what happens in bigger concert venues. The first installment, featuring folk artist Laura Gibson, went up on YouTube. Today, the concerts are posted on the NPR site with a writeup and credits, as well as YouTube, where NPR Music has 11 million followers. NPR Music also clips installments on Instagram, where it has 3 million followers. 

    In the early days, NPR staff reached out to touring bands to secure bookings. Acts coming through DC could often be cajoled into filming an installment before heading out to their venues for that night’s sound check. Now, musicians come to DC just for the chance to record in NPR’s offices. 

    “We don’t have to worry about tours anymore,” Carter says. “Labels and artists are willing to come in solely for a Tiny Desk performance. They understand the impact that a really good Tiny Desk concert can have on an artist’s career.”

    Early on, the stripped-down nature of the Tiny Desk—artists can’t use any audio processing or voice modulation—lent itself to rock, folk, and indie acts. But a 2014 concert with T-Pain, in which the famously autotune-heavy singer unveiled an impressive set of pipes, showed how artists from a broader array of genres could shine behind the Tiny Desk. 

    “Everyone knows at this point that they’re going to have to do something different in our space,” Carter says. “It’s a bigger ask for hip-hop acts and electronic acts, but most artists now understand how important it can be if they nail it.”

    Carter highlights rapper Doechii as an artist who overhauled her sound for her Tiny Desk concert in December. Doechii’s all-female backing band used trumpet, saxophone, guitar, and bass to transform songs from her mixtape Alligator Bites Never Heal for the live setting. “If you listen to the recorded version of her music, it’s nothing like what you saw in that Tiny Desk,” Carter says. 

    Clips of Doechii’s Tiny Desk virtuosity lit up social media, introducing the ‘swamp princess’ to new fans. The concert even inspired a viral parody, with writer-director-comedian Gus Heagary pretending to be an NPR staffer watching the performance.   

    Reimagining Old Favorites

    It isn’t just emerging acts that totally revamp their sound for a Tiny Desk opportunity. Established artists like Usher, Justin Timberlake, and Cypress Hill have followed T-Pain’s lead and used NPR’s offices to showcase reimagined versions of some of their most popular songs. When Juvenile recorded his installment in June 2023, he was backed by horns and saxophones, a violin and cello, and John Batiste on melodica. The New Orleans rapper played an acoustic version of “Back That Azz Up” twice at the audience’s request—the first encore in the series’ history. 

    “I love what has happened with hip hop,” Carter says. He explains that artists now approach the concert with the mindset: ‘I have to really rethink what I’ve been doing for however long I’ve been doing it, and present it in a whole new way.” 

    Tiny Desk has also helped musicians like Juvenile, gospel artist Marvin Sapp, and percussionist Sheila E to reach new audiences while reminding listeners they’re still making music. “We’re helping artists to re-emerge,” Carter says, “tapping into legacy acts and evergreen artistsbreathe new life into their careers.”

    In many ways, Tiny Desk now occupies a niche once filled by MTV Unplugged—but for the generation that has replaced cable with YouTube and streaming.  

    “Maybe 10, 15, 20 years ago, all of our favorite artists had this watershed moment in terms of a live performance,” Carter says. “Back in the day it was MTV Unplugged. SNL is still doing their thing. But when you think about the generation now that lives on YouTube, some of these Tiny Desk performances are going to be the milestone that people point to when it comes to live performances.”

    Building a Diverse Audience

    When Carter talks about Tiny Desk concerts reaching a new generation of listeners, it’s not conjecture. He notes that the NPR Music YouTube channel’s 11 million subscribers are “as young and diverse as it gets. It’s almost half people of colormuch younger than the audience that listens to NPR on air, which is an audience NPR has been trying to tap for a long time,” he says. 

    That diversity informs some of the special series that Tiny Desk produces. The Juvenile video was part of Carter’s second run of concerts recorded for Black Music Month, in June. Ca7riel & Paco Amoroso’s video was tied to El Tiny, a Latin-focused series that debuts during Latin Heritage Monthand is programmed by Tiny Desk producer and Alt.Latino host AnaMaria Sayer. 

    Ca7riel & Paco Amoroso’s tour manager, Izquierdo, has worked with artists featured in the series before. He says Tiny Desk is crucial for Latin American artists trying to break through. “I’ve realized that for U.S. radio, Latin music benefits from Tiny Desk,” he says.

    The Tiny Desk audience’s broad demographics are also increasingly reflected in its broader programming. Bad Bunny’s April installment took his reggaeton-inspired songs from recent album Debi Tirar Mas Fotos to their acoustic roots, using an array of traditional Puerto Rican, Latin American, and Caribbean instruments, such as the cuatro puertorriqueño, tiple, güicharo, and bongos.  “audience informs a whole lot of what we do,” Carter says. I get so many pointers from YouTube comments like ‘Have you heard of this artist?’ We’re watching all that stuff because it helps us stay sharp.”

    Tiny Desk heard round the world

    With a strong global audience, Tiny Desk has been expanding into Asia. In 2023, NPR struck a licensing deal with South Korean Telecom LG U+ and production company Something Special to produce Tiny Desk Korea for television. Last year, NPR inked a deal with the Japan Broadcasting Corporationto launch Tiny Desk Concerts Japan. “We’re really expanding in terms of global reach,” Carter says. 

    Here in the States, Carter and Sayer recently launched Tiny Desk Radio, a series that will revisit some of the series’ notable installments, sharing behind-the-scenes stories from their productions and playing the audio from the concerts “Our engineers put a lot of time and effort into making sure that we sound great,” Carter says. “I hear it a lot—people tell me they prefer an artist’s Tiny Desk over anything.”

    That’s something Ca7riel & Paco Amoroso clearly have on their mind as they navigate the Tiny Desk effect and a new level of recognition. The duo released an EP in February, Papota, which features four new songs, plus the recorded versions of their pared-down Tiny Desk performances. They also released a short film that recreates their Tiny Desk performance—this time in a Buenos Aires diner.

    One of the themes of the EP is the pair wrestling with the implications of their viral success. On the song Impostor, Ca7riel asks “¿Y ahora que vamos hacer?/El tiny desk me jodio”It’s an overstatement, but an acknowledgment that the path they’re now on ran directly through the NPR offices. 
    #how #nprs #tiny #desk #became
    How NPR’s Tiny Desk became the biggest stage in music
    Until last October, Argentinian musical duo Ca7riel & Paco Amoroso were more or less a regional act. Known for their experimental blend of Latin trap, pop, and rap, the pair had a fanbase, but still weren’t cracking more than 3,000 daily streams across services like Spotify, Apple Music, and YouTube. Within a week, they shot up 4,700%—hitting 222,000 daily streams—according to exclusive data firm Luminate, which powers the Billboard charts. Suddenly Ca7riel & Paco Amoroso were global pop stars.  What changed? On Oct. 4, the pair were featured in a Tiny Desk Concert, part of NPR’s 17-year-old video series featuring musicians performing stripped-down sets behind an office desk in the cramped Washington, D.C. headquarters of the public broadcaster.  In the concert video, the artists play five songs from their debut album Baño Maria, which came out last April. Paco’s raspy voice emerges from underneath a puffy blue trapper hat while Ca7riel sports an over-the-top pout and a vest made of stitched-together heart-shaped plush toys. The pair sing entirely in Spanish, backed by their Argentinian bandmatesand an American horn section. The duo’s performance quickly took off across the internet. Within five days, it had racked up more than 1.5 million views on YouTube, and hit 11 million in little more than a month. It also reverberated across social media: the NPR Music Instagram post garnering nearly 900,000 likes, and TikToks clips garnered hundreds of thousands of views.  In a year that featured Tiny Desk performances from buzzy stars like Chappell Roan and Sabrina Carpenter, as well as established acts like Chaka Khan and Nelly Furtado, Ca7riel & Paco Amoroso’s concert was the most-watched of 2024. It currently sits at 36 million views.  That virality translated to an influx of bookings for the duo, including a performance at Coachella in April, and upcoming slots at Glastonbury in June, FujiRock Japan in July, and Lollapalooza and Outside Lands in August. Ca7riel & Paco Amoroso’s global tour includes sold-out dates at Mexico’s 20,000-capacity Palacio de los Deportes and Chile’s 14,000-seat Movistar Areana—and was previewed by an appearance on The Tonight Show Starring Jimmy Fallon in April.  “Through Tiny Desk, we’ve noticed media approaching us, promoters being very interested in offering their spaces and festivals, and many media outlets opening doors to show us to the world,” says Jonathan Izquierdo, the band’s Spain-based tour manager who began working with the duo shortly after the Tiny Desk Concert debuted. “We’ve managed to sell out summer arena shows in record time and we’re constantly adding new concerts. Promoters are knocking on our doors to get the Tiny Desk effect.” Bobby CarterTiny Desk, Big Influence The Tiny Desk effect is something Bobby Carter, NPR Tiny Desk host and series producer, has seen firsthand. Carter has been at NPR for 25 years, including the past 11 on the Tiny Desk team. He took the reins when Bob Boilen, the longtime All Songs Considered host who launched Tiny Desk in 2008, retired in 2023.  The series—which now has more than 1,200 videos—began as an internet-first way for Boilen to showcase performances from musicians that were more intimate than what happens in bigger concert venues. The first installment, featuring folk artist Laura Gibson, went up on YouTube. Today, the concerts are posted on the NPR site with a writeup and credits, as well as YouTube, where NPR Music has 11 million followers. NPR Music also clips installments on Instagram, where it has 3 million followers.  In the early days, NPR staff reached out to touring bands to secure bookings. Acts coming through DC could often be cajoled into filming an installment before heading out to their venues for that night’s sound check. Now, musicians come to DC just for the chance to record in NPR’s offices.  “We don’t have to worry about tours anymore,” Carter says. “Labels and artists are willing to come in solely for a Tiny Desk performance. They understand the impact that a really good Tiny Desk concert can have on an artist’s career.” Early on, the stripped-down nature of the Tiny Desk—artists can’t use any audio processing or voice modulation—lent itself to rock, folk, and indie acts. But a 2014 concert with T-Pain, in which the famously autotune-heavy singer unveiled an impressive set of pipes, showed how artists from a broader array of genres could shine behind the Tiny Desk.  “Everyone knows at this point that they’re going to have to do something different in our space,” Carter says. “It’s a bigger ask for hip-hop acts and electronic acts, but most artists now understand how important it can be if they nail it.” Carter highlights rapper Doechii as an artist who overhauled her sound for her Tiny Desk concert in December. Doechii’s all-female backing band used trumpet, saxophone, guitar, and bass to transform songs from her mixtape Alligator Bites Never Heal for the live setting. “If you listen to the recorded version of her music, it’s nothing like what you saw in that Tiny Desk,” Carter says.  Clips of Doechii’s Tiny Desk virtuosity lit up social media, introducing the ‘swamp princess’ to new fans. The concert even inspired a viral parody, with writer-director-comedian Gus Heagary pretending to be an NPR staffer watching the performance.    Reimagining Old Favorites It isn’t just emerging acts that totally revamp their sound for a Tiny Desk opportunity. Established artists like Usher, Justin Timberlake, and Cypress Hill have followed T-Pain’s lead and used NPR’s offices to showcase reimagined versions of some of their most popular songs. When Juvenile recorded his installment in June 2023, he was backed by horns and saxophones, a violin and cello, and John Batiste on melodica. The New Orleans rapper played an acoustic version of “Back That Azz Up” twice at the audience’s request—the first encore in the series’ history.  “I love what has happened with hip hop,” Carter says. He explains that artists now approach the concert with the mindset: ‘I have to really rethink what I’ve been doing for however long I’ve been doing it, and present it in a whole new way.”  Tiny Desk has also helped musicians like Juvenile, gospel artist Marvin Sapp, and percussionist Sheila E to reach new audiences while reminding listeners they’re still making music. “We’re helping artists to re-emerge,” Carter says, “tapping into legacy acts and evergreen artistsbreathe new life into their careers.” In many ways, Tiny Desk now occupies a niche once filled by MTV Unplugged—but for the generation that has replaced cable with YouTube and streaming.   “Maybe 10, 15, 20 years ago, all of our favorite artists had this watershed moment in terms of a live performance,” Carter says. “Back in the day it was MTV Unplugged. SNL is still doing their thing. But when you think about the generation now that lives on YouTube, some of these Tiny Desk performances are going to be the milestone that people point to when it comes to live performances.” Building a Diverse Audience When Carter talks about Tiny Desk concerts reaching a new generation of listeners, it’s not conjecture. He notes that the NPR Music YouTube channel’s 11 million subscribers are “as young and diverse as it gets. It’s almost half people of colormuch younger than the audience that listens to NPR on air, which is an audience NPR has been trying to tap for a long time,” he says.  That diversity informs some of the special series that Tiny Desk produces. The Juvenile video was part of Carter’s second run of concerts recorded for Black Music Month, in June. Ca7riel & Paco Amoroso’s video was tied to El Tiny, a Latin-focused series that debuts during Latin Heritage Monthand is programmed by Tiny Desk producer and Alt.Latino host AnaMaria Sayer.  Ca7riel & Paco Amoroso’s tour manager, Izquierdo, has worked with artists featured in the series before. He says Tiny Desk is crucial for Latin American artists trying to break through. “I’ve realized that for U.S. radio, Latin music benefits from Tiny Desk,” he says. The Tiny Desk audience’s broad demographics are also increasingly reflected in its broader programming. Bad Bunny’s April installment took his reggaeton-inspired songs from recent album Debi Tirar Mas Fotos to their acoustic roots, using an array of traditional Puerto Rican, Latin American, and Caribbean instruments, such as the cuatro puertorriqueño, tiple, güicharo, and bongos.  “audience informs a whole lot of what we do,” Carter says. I get so many pointers from YouTube comments like ‘Have you heard of this artist?’ We’re watching all that stuff because it helps us stay sharp.” Tiny Desk heard round the world With a strong global audience, Tiny Desk has been expanding into Asia. In 2023, NPR struck a licensing deal with South Korean Telecom LG U+ and production company Something Special to produce Tiny Desk Korea for television. Last year, NPR inked a deal with the Japan Broadcasting Corporationto launch Tiny Desk Concerts Japan. “We’re really expanding in terms of global reach,” Carter says.  Here in the States, Carter and Sayer recently launched Tiny Desk Radio, a series that will revisit some of the series’ notable installments, sharing behind-the-scenes stories from their productions and playing the audio from the concerts “Our engineers put a lot of time and effort into making sure that we sound great,” Carter says. “I hear it a lot—people tell me they prefer an artist’s Tiny Desk over anything.” That’s something Ca7riel & Paco Amoroso clearly have on their mind as they navigate the Tiny Desk effect and a new level of recognition. The duo released an EP in February, Papota, which features four new songs, plus the recorded versions of their pared-down Tiny Desk performances. They also released a short film that recreates their Tiny Desk performance—this time in a Buenos Aires diner. One of the themes of the EP is the pair wrestling with the implications of their viral success. On the song Impostor, Ca7riel asks “¿Y ahora que vamos hacer?/El tiny desk me jodio”It’s an overstatement, but an acknowledgment that the path they’re now on ran directly through the NPR offices.  #how #nprs #tiny #desk #became
    WWW.FASTCOMPANY.COM
    How NPR’s Tiny Desk became the biggest stage in music
    Until last October, Argentinian musical duo Ca7riel & Paco Amoroso were more or less a regional act. Known for their experimental blend of Latin trap, pop, and rap, the pair had a fanbase, but still weren’t cracking more than 3,000 daily streams across services like Spotify, Apple Music, and YouTube. Within a week, they shot up 4,700%—hitting 222,000 daily streams—according to exclusive data firm Luminate, which powers the Billboard charts. Suddenly Ca7riel & Paco Amoroso were global pop stars.  What changed? On Oct. 4, the pair were featured in a Tiny Desk Concert, part of NPR’s 17-year-old video series featuring musicians performing stripped-down sets behind an office desk in the cramped Washington, D.C. headquarters of the public broadcaster.  In the concert video, the artists play five songs from their debut album Baño Maria, which came out last April. Paco’s raspy voice emerges from underneath a puffy blue trapper hat while Ca7riel sports an over-the-top pout and a vest made of stitched-together heart-shaped plush toys. The pair sing entirely in Spanish, backed by their Argentinian bandmates (sporting shirts screenprinted with their visas) and an American horn section. The duo’s performance quickly took off across the internet. Within five days, it had racked up more than 1.5 million views on YouTube, and hit 11 million in little more than a month. It also reverberated across social media: the NPR Music Instagram post garnering nearly 900,000 likes, and TikToks clips garnered hundreds of thousands of views.  In a year that featured Tiny Desk performances from buzzy stars like Chappell Roan and Sabrina Carpenter, as well as established acts like Chaka Khan and Nelly Furtado, Ca7riel & Paco Amoroso’s concert was the most-watched of 2024. It currently sits at 36 million views.  That virality translated to an influx of bookings for the duo, including a performance at Coachella in April, and upcoming slots at Glastonbury in June, FujiRock Japan in July, and Lollapalooza and Outside Lands in August. Ca7riel & Paco Amoroso’s global tour includes sold-out dates at Mexico’s 20,000-capacity Palacio de los Deportes and Chile’s 14,000-seat Movistar Areana—and was previewed by an appearance on The Tonight Show Starring Jimmy Fallon in April.  “Through Tiny Desk, we’ve noticed media approaching us, promoters being very interested in offering their spaces and festivals, and many media outlets opening doors to show us to the world,” says Jonathan Izquierdo, the band’s Spain-based tour manager who began working with the duo shortly after the Tiny Desk Concert debuted. “We’ve managed to sell out summer arena shows in record time and we’re constantly adding new concerts. Promoters are knocking on our doors to get the Tiny Desk effect.” Bobby Carter [Photo: Fenn Paider/courtesy NPR] Tiny Desk, Big Influence The Tiny Desk effect is something Bobby Carter, NPR Tiny Desk host and series producer, has seen firsthand. Carter has been at NPR for 25 years, including the past 11 on the Tiny Desk team. He took the reins when Bob Boilen, the longtime All Songs Considered host who launched Tiny Desk in 2008, retired in 2023.  The series—which now has more than 1,200 videos—began as an internet-first way for Boilen to showcase performances from musicians that were more intimate than what happens in bigger concert venues. The first installment, featuring folk artist Laura Gibson, went up on YouTube. Today, the concerts are posted on the NPR site with a writeup and credits, as well as YouTube, where NPR Music has 11 million followers. NPR Music also clips installments on Instagram, where it has 3 million followers.  In the early days, NPR staff reached out to touring bands to secure bookings. Acts coming through DC could often be cajoled into filming an installment before heading out to their venues for that night’s sound check. Now, musicians come to DC just for the chance to record in NPR’s offices.  “We don’t have to worry about tours anymore,” Carter says. “Labels and artists are willing to come in solely for a Tiny Desk performance. They understand the impact that a really good Tiny Desk concert can have on an artist’s career.” Early on, the stripped-down nature of the Tiny Desk—artists can’t use any audio processing or voice modulation—lent itself to rock, folk, and indie acts. But a 2014 concert with T-Pain, in which the famously autotune-heavy singer unveiled an impressive set of pipes, showed how artists from a broader array of genres could shine behind the Tiny Desk.  “Everyone knows at this point that they’re going to have to do something different in our space,” Carter says. “It’s a bigger ask for hip-hop acts and electronic acts, but most artists now understand how important it can be if they nail it.” Carter highlights rapper Doechii as an artist who overhauled her sound for her Tiny Desk concert in December. Doechii’s all-female backing band used trumpet, saxophone, guitar, and bass to transform songs from her mixtape Alligator Bites Never Heal for the live setting. “If you listen to the recorded version of her music, it’s nothing like what you saw in that Tiny Desk,” Carter says.  Clips of Doechii’s Tiny Desk virtuosity lit up social media, introducing the ‘swamp princess’ to new fans. The concert even inspired a viral parody, with writer-director-comedian Gus Heagary pretending to be an NPR staffer watching the performance.    Reimagining Old Favorites It isn’t just emerging acts that totally revamp their sound for a Tiny Desk opportunity. Established artists like Usher, Justin Timberlake, and Cypress Hill have followed T-Pain’s lead and used NPR’s offices to showcase reimagined versions of some of their most popular songs. When Juvenile recorded his installment in June 2023, he was backed by horns and saxophones, a violin and cello, and John Batiste on melodica. The New Orleans rapper played an acoustic version of “Back That Azz Up” twice at the audience’s request—the first encore in the series’ history.  “I love what has happened with hip hop [on Tiny Desk],” Carter says. He explains that artists now approach the concert with the mindset: ‘I have to really rethink what I’ve been doing for however long I’ve been doing it, and present it in a whole new way.”  Tiny Desk has also helped musicians like Juvenile, gospel artist Marvin Sapp, and percussionist Sheila E to reach new audiences while reminding listeners they’re still making music. “We’re helping artists to re-emerge,” Carter says, “tapping into legacy acts and evergreen artists [to help] breathe new life into their careers.” In many ways, Tiny Desk now occupies a niche once filled by MTV Unplugged—but for the generation that has replaced cable with YouTube and streaming.   “Maybe 10, 15, 20 years ago, all of our favorite artists had this watershed moment in terms of a live performance,” Carter says. “Back in the day it was MTV Unplugged. SNL is still doing their thing. But when you think about the generation now that lives on YouTube, some of these Tiny Desk performances are going to be the milestone that people point to when it comes to live performances.” Building a Diverse Audience When Carter talks about Tiny Desk concerts reaching a new generation of listeners, it’s not conjecture. He notes that the NPR Music YouTube channel’s 11 million subscribers are “as young and diverse as it gets. It’s almost half people of color [and] much younger than the audience that listens to NPR on air, which is an audience NPR has been trying to tap for a long time,” he says.  That diversity informs some of the special series that Tiny Desk produces. The Juvenile video was part of Carter’s second run of concerts recorded for Black Music Month, in June. Ca7riel & Paco Amoroso’s video was tied to El Tiny, a Latin-focused series that debuts during Latin Heritage Month (from mid September to mid October) and is programmed by Tiny Desk producer and Alt.Latino host AnaMaria Sayer.  Ca7riel & Paco Amoroso’s tour manager, Izquierdo, has worked with artists featured in the series before. He says Tiny Desk is crucial for Latin American artists trying to break through. “I’ve realized that for U.S. radio, Latin music benefits from Tiny Desk,” he says. The Tiny Desk audience’s broad demographics are also increasingly reflected in its broader programming. Bad Bunny’s April installment took his reggaeton-inspired songs from recent album Debi Tirar Mas Fotos to their acoustic roots, using an array of traditional Puerto Rican, Latin American, and Caribbean instruments, such as the cuatro puertorriqueño, tiple, güicharo, and bongos.  “[Our] audience informs a whole lot of what we do,” Carter says. I get so many pointers from YouTube comments like ‘Have you heard of this artist?’ We’re watching all that stuff because it helps us stay sharp.” Tiny Desk heard round the world With a strong global audience, Tiny Desk has been expanding into Asia. In 2023, NPR struck a licensing deal with South Korean Telecom LG U+ and production company Something Special to produce Tiny Desk Korea for television. Last year, NPR inked a deal with the Japan Broadcasting Corporation (NHK) to launch Tiny Desk Concerts Japan. “We’re really expanding in terms of global reach,” Carter says.  Here in the States, Carter and Sayer recently launched Tiny Desk Radio, a series that will revisit some of the series’ notable installments, sharing behind-the-scenes stories from their productions and playing the audio from the concerts “Our engineers put a lot of time and effort into making sure that we sound great,” Carter says. “I hear it a lot—people tell me they prefer an artist’s Tiny Desk over anything.” That’s something Ca7riel & Paco Amoroso clearly have on their mind as they navigate the Tiny Desk effect and a new level of recognition (their daily streams haven’t dipped below 50,000 a day since the beginning of the year). The duo released an EP in February, Papota, which features four new songs, plus the recorded versions of their pared-down Tiny Desk performances. They also released a short film that recreates their Tiny Desk performance—this time in a Buenos Aires diner. One of the themes of the EP is the pair wrestling with the implications of their viral success. On the song Impostor, Ca7riel asks “¿Y ahora que vamos hacer?/El tiny desk me jodio” (What do we do now? Tiny Desk fucked me up.) It’s an overstatement, but an acknowledgment that the path they’re now on ran directly through the NPR offices. 
    0 Comentários 0 Compartilhamentos
  • The Best Multiplayer Video Games for 2025

    Sometimes you want to play alone, whether your game of choice is a relaxing solitaire session or an engrossing, cinematic campaign. We get that. Still, some of the best gaming-related experiences come from moments shared with other people. After all, an excellent multiplayer mode makes a video game endlessly replayable and enables good times with local friends or strangers across the country—as long as the servers stay up. Our list of the best multiplayer games casts a wide net that includes console and PC games, competitive and cooperative titles, casual board games and serious esports fare, and, of course, battle royales, shooters, and fighters. If you're interested in playing a game with at least one other person, you'll find something that catches your eye here. These are the best multiplayer video games that you and your friends should play right now.

    Apex Legends

    From the ashes of Titanfall rose the best battle royale game. Respawn’s Apex Legends combines unbelievably fluid movement with impeccable gunplay and innovative team communication features. Each character’s unique abilities open strategic options on the expansive battlefield.

    Among Us

    Among Us4.0 Excellent

    Among Us is more of a social experiment than a game. You and your friends play as crewmates attempting to repair a spaceship, but some players are deadly impostors who are picking off others. Constant lying and manipulation turn even the friendliest relationships into pure paranoia. 

    Clubhouse Games

    Clubhouse Games: 51 Worldwide Classics4.0 Excellent

    This compilation contains more than 50 classic games that have stood the test of time, including bowling, backgammon, and billiards. You can have fun with friends or family, but beware getting so heated that you'll never want to speak with them again. Online play ensures the goodtimes aren't limited to your immediate vicinity.

    Counter-Strike 2

    Counter-Strike: Global Offensive

    4.0 Excellent

    Counter-Strike is a founding father of multiplayer online shooters, and Counter-Strike 2 continues the legacy. In this long-awaited update to Counter-Strike: Global Offensive, you’ll find a massive community always ready to hop into Terrorist versus Counter-Terrorist tactical team action. Plus, it’s free!

    Death Stranding

    Death Stranding: Director's Cut4.0 Excellent

    Hideo Kojima’s freaky odyssey about time rain and babies in jars is also one of the most fascinating and unconventional multiplayer games in recent memory. As you traverse the harsh wasteland, you can leave behind useful items, such as ladders and reports, that other players can use in their sessions.

    Destiny 2

    Destiny 2Destiny 2 is the looter-shooter that gives other looter-shooters envy. You gather the shiniest guns, the sickest armor, and show off your gear in front of fellow Guardians. Party-up and shoot your way through alien enemies and strongholds with Bungie’s perfect first-person shooter controls. The first taste is free, and regular, new content releases give you many reasons to keep gunning.
    Destiny 2review

    Diablo IV

    4.0 Excellent

    No action-RPG out-Diablos Diablo IV, a title that expands the familiar loot-grinding mechanics with massive, demon-filled zones. In terms of character builds,
    Diablo IV review

    Dota 2

    Dota 24.5 Excellent

    What began as a mere mod has since become one of the most popular esports in the world. Dota 2 sets the standard for the MOBA genre, that strange hybrid between real-time strategy and team sports. New heroes give players constantly changing choices to consider. If you put in the effort to get really good at this game, the sky's the limit. 

    Fatal Fury: City of the Wolves

    Fatal Fury: City of the Wolves4.0 Excellent

    The King of Fighters series is great, but Fatal Fury: City of the Wolves resurrects the SNK fighter that started it all. Familiar faces like Terry Bogard and Mai Shiranui battle real-life guest characters like DJ Salavatore Gannaci and soccer star Cristiano Ronaldo in this excellent take on fundamental, 2D fisticuffs. Rich mechanics add depth to both offensive and defensive play, while comic book-inspired graphics give brawls a distinct visual identity. Crossplay multiplayer shines with rollback netcode.
    Fatal Fury: City of the Wolvesreview

    Fortnite

    Fortnite3.5 Good

    Do you really need someone else to tell you about Fortnite? Originally a free battle royale mode for a failed multiplayer game, Fortnite became an absolute phenomenon. Every day, millions of children leap from the in-game battle bus to shoot each other and build elaborate structures, while dressed as their favorite brands. You can also hang out and watch concerts ordocumentaries on social issues.  

    Forza Horizon 5

    Forza Horizon 54.5 Excellent

    Forza Horizon 5 appeals equally to serious automobile racing enthusiasts and anyone who just wants to drive aimlessly through beautifully rendered Mexican landscapes. Although largely similar to past entries, the new EventLab lets you create clever, custom courses to enjoy with friends.

    Halo Infinite

    Halo Infinite4.5 Excellent

    Halo single-handedly saved the Xbox, and proved that multiplayer shooters could thrive on home consoles. Halo Infinite doesn’t just reinvent the single-player campaign, it continues Halo’s history of excellent multiplayer modes, from capture the flag to random weapon fiestas. Plus, you can play for free, so finish the fight.
    Halo Infinitereview

    Jackbox Party Pack

    The Jackbox Party Pack 8The annual Jackbox Party Pack games consistently deliver the most hilarious social multiplayer experiences you’ll ever play. Design wacky t-shirts, come up with witty quips, and try to figure out which friend is faking it. Anyone can play, as long as they have a phone. With unique streaming features, even your audience can join the party.

    League of Legends

    League of Legends4.5 Excellent

    Free from any previous mod baggage, League of Legends is arguably the more accessible game when it comes to the MOBA heavy hitters. Still, it takes skill to master every champion and lead your team to victory. The League of Legends universe is expanding into other game genres and Netflix shows, so now’s the time to get caught up. 
    League of Legendsreview

    It Takes Two

    In many ways, marriage is the ultimate multiplayer game. It Takes Two is a cooperative adventure that tasks two people with controlling a couple as they complete wacky challenges to repair their strained relationship. You’ll never know true love until you and your partner escape a giant cuckoo clock together.

    The King of Fighters XV

    The King of Fighters XV4.0 Excellent

    For finely tuned 2D fighting, look no further than The King of Fighters XV. Building off intriguing ideas introduced in previous entries, KOF XV gives you a massive character roster and an expressive, creative fighting system. Tournament features, multiplayer party modes, and rollback netcode make this one of the series' best entries.

    Mario Kart 8 Deluxe

    Mario Kart 8 Deluxe4.5 Excellent

    Mario Kart 8 Deluxe is absolutely everything you could want from Nintendo’s hugely popular kart racing series. It features gorgeous visuals, inventive tracks, and a revamped battle mode. In fact, Nintendo is still selling new courses, years after the game's 2017 debut. Mario Kart 8 Deluxe is so spectacular not even the blue shell can stop it. 

    Minecraft

    Minecraft4.5 Excellent

    Minecraft gives young people an unparalleled sense of freedom as they explore and build worlds, brick by brick. In fact, multiple builders can join the same game world for cooperative mining and crafting. Take it a step further by setting up your own Minecraft server, so you and friends can construct a private paradise. 

    Monster Hunter Rise

    Monster Hunter Rise4.5 Excellent

    Monster Hunter Rise finally makes hardened haters see the glory of Capcom’s monster-slaying series. You can craft new weapons and armors by defeating a monster menagerie inspired by Japanese mythology or swing through the air with new wirebug techniques. No beasts will stand in your way as you go beast hunting with friends.

    Mortal Kombat 1

    Mortal Kombat 14.5 Excellent

    With Mortal Kombat 1, the famously bloody fighting game finally breaks free of its infamously stiff gameplay mechanics. Finishing foes has never been more fun thanks to high-flying air combos and custom tag-team Kameo fighter attacks. This reboot is as entertaining to play with friends as it is to watch with horrified onlookers.
    Mortal Kombat 1review

    Rocket League

    Rocket League4.5 Excellent

    “Cars playing soccer” is such a beautiful premise for an arcade sports game, and Rocket League perfectly pulls it off. Sure, you can just put the pedal to the metal and bash into the ball, hoping it goes into the goal. But the high-flying physics system creates enough depth for sensational tests of skill. The free-to-play season structure means you’ll always have a reason to return. 
    Rocket Leaguereview

    Splatoon 3

    Splatoon 34.0 Excellent

    Only Nintendo could take the well-worn shooter genre and turn it into a game about punky squid kids squirting ink at each other. By making battles more about covering turf than blasting opponents, Splatoon 3 is a friendlier and more accessible shooter. However, you’ll need to stay on your toes with so many unique weapons and traversal options.
    Splatoon 3review

    StarCraft II

    StarCraft II: Legacy of the Void5.0 Outstanding

    StarCraft II is the best strategy game since chess. Whether you play as Terran, Zerg, or Protoss armies, you have access to perfectly balanced units for overcoming any opponent during real-time clashes. The StarCraft II trilogy even introduces free, cooperate multiplayer modes, so veterans can introduce newcomers to the fight.

    Street Fighter 6

    Street Fighter 65.0 Outstanding

    Street Fighter 6 is worthy of its iconic name. With its bold new style, expansive new modes, exciting new roster, and competitive new gameplay systems, Street Fighter battles are more hype than ever. It's a multiplayer gaming masterpiece.
    Street Fighter 6review

    Streets of Rage 4

    Streets of Rage 44.0 Excellent

    Streets of Rage 4 breathes new life into the aging beat ‘em up genre thanks to complex combat and stunning illustrated graphics. If smacking goons in solo fashion gets boring, team up with a friend for chaotic co-op action. You can even unlock retro versions of classic characters.

    Super Smash Bros. Ultimate

    Super Smash Bros. Ultimate4.5 Excellent

    Super Smash Bros. Ultimate combines countless characters, stages, modes, and music tracks to create the most incredible video game crossover of all time. It’s also a phenomenal platform-fighting game, speeding up the addictive combat and rebalancing advanced techniques. No matter how seriously you take it, no video game can scratch that satisfying multiplayer itch like Super Smash Bros. 

    Tekken 8

    Tekken 8Tekken 8 is the latest and greatest entry in the venerable 3D fighting game franchise. Prove your worth at the King of Iron Fist Tournament by nimbly sidestepping, laying down painful combos, and activating new Heat Smash attacks. For the true Tekken multiplayer experience, fight your dad and throw him down a volcano.

    Tetris Effect: Connected

    Tetris Effect: Connected4.5 Excellent

    Tetris Effect: Connected makes the perfect puzzle game even better. Alongside traditional competitive Tetris multiplayer modes, Effect lets you team up for cooperative “Connected” journeys where you and your partners clear lines on the same massive board. Combine that with trance-inducing audiovisual stimuli, and you’ll never look at blocks the same way again. 
    #best #multiplayer #video #games
    The Best Multiplayer Video Games for 2025
    Sometimes you want to play alone, whether your game of choice is a relaxing solitaire session or an engrossing, cinematic campaign. We get that. Still, some of the best gaming-related experiences come from moments shared with other people. After all, an excellent multiplayer mode makes a video game endlessly replayable and enables good times with local friends or strangers across the country—as long as the servers stay up. Our list of the best multiplayer games casts a wide net that includes console and PC games, competitive and cooperative titles, casual board games and serious esports fare, and, of course, battle royales, shooters, and fighters. If you're interested in playing a game with at least one other person, you'll find something that catches your eye here. These are the best multiplayer video games that you and your friends should play right now. Apex Legends From the ashes of Titanfall rose the best battle royale game. Respawn’s Apex Legends combines unbelievably fluid movement with impeccable gunplay and innovative team communication features. Each character’s unique abilities open strategic options on the expansive battlefield. Among Us Among Us4.0 Excellent Among Us is more of a social experiment than a game. You and your friends play as crewmates attempting to repair a spaceship, but some players are deadly impostors who are picking off others. Constant lying and manipulation turn even the friendliest relationships into pure paranoia.  Clubhouse Games Clubhouse Games: 51 Worldwide Classics4.0 Excellent This compilation contains more than 50 classic games that have stood the test of time, including bowling, backgammon, and billiards. You can have fun with friends or family, but beware getting so heated that you'll never want to speak with them again. Online play ensures the goodtimes aren't limited to your immediate vicinity. Counter-Strike 2 Counter-Strike: Global Offensive 4.0 Excellent Counter-Strike is a founding father of multiplayer online shooters, and Counter-Strike 2 continues the legacy. In this long-awaited update to Counter-Strike: Global Offensive, you’ll find a massive community always ready to hop into Terrorist versus Counter-Terrorist tactical team action. Plus, it’s free! Death Stranding Death Stranding: Director's Cut4.0 Excellent Hideo Kojima’s freaky odyssey about time rain and babies in jars is also one of the most fascinating and unconventional multiplayer games in recent memory. As you traverse the harsh wasteland, you can leave behind useful items, such as ladders and reports, that other players can use in their sessions. Destiny 2 Destiny 2Destiny 2 is the looter-shooter that gives other looter-shooters envy. You gather the shiniest guns, the sickest armor, and show off your gear in front of fellow Guardians. Party-up and shoot your way through alien enemies and strongholds with Bungie’s perfect first-person shooter controls. The first taste is free, and regular, new content releases give you many reasons to keep gunning. Destiny 2review Diablo IV 4.0 Excellent No action-RPG out-Diablos Diablo IV, a title that expands the familiar loot-grinding mechanics with massive, demon-filled zones. In terms of character builds, Diablo IV review Dota 2 Dota 24.5 Excellent What began as a mere mod has since become one of the most popular esports in the world. Dota 2 sets the standard for the MOBA genre, that strange hybrid between real-time strategy and team sports. New heroes give players constantly changing choices to consider. If you put in the effort to get really good at this game, the sky's the limit.  Fatal Fury: City of the Wolves Fatal Fury: City of the Wolves4.0 Excellent The King of Fighters series is great, but Fatal Fury: City of the Wolves resurrects the SNK fighter that started it all. Familiar faces like Terry Bogard and Mai Shiranui battle real-life guest characters like DJ Salavatore Gannaci and soccer star Cristiano Ronaldo in this excellent take on fundamental, 2D fisticuffs. Rich mechanics add depth to both offensive and defensive play, while comic book-inspired graphics give brawls a distinct visual identity. Crossplay multiplayer shines with rollback netcode. Fatal Fury: City of the Wolvesreview Fortnite Fortnite3.5 Good Do you really need someone else to tell you about Fortnite? Originally a free battle royale mode for a failed multiplayer game, Fortnite became an absolute phenomenon. Every day, millions of children leap from the in-game battle bus to shoot each other and build elaborate structures, while dressed as their favorite brands. You can also hang out and watch concerts ordocumentaries on social issues.   Forza Horizon 5 Forza Horizon 54.5 Excellent Forza Horizon 5 appeals equally to serious automobile racing enthusiasts and anyone who just wants to drive aimlessly through beautifully rendered Mexican landscapes. Although largely similar to past entries, the new EventLab lets you create clever, custom courses to enjoy with friends. Halo Infinite Halo Infinite4.5 Excellent Halo single-handedly saved the Xbox, and proved that multiplayer shooters could thrive on home consoles. Halo Infinite doesn’t just reinvent the single-player campaign, it continues Halo’s history of excellent multiplayer modes, from capture the flag to random weapon fiestas. Plus, you can play for free, so finish the fight. Halo Infinitereview Jackbox Party Pack The Jackbox Party Pack 8The annual Jackbox Party Pack games consistently deliver the most hilarious social multiplayer experiences you’ll ever play. Design wacky t-shirts, come up with witty quips, and try to figure out which friend is faking it. Anyone can play, as long as they have a phone. With unique streaming features, even your audience can join the party. League of Legends League of Legends4.5 Excellent Free from any previous mod baggage, League of Legends is arguably the more accessible game when it comes to the MOBA heavy hitters. Still, it takes skill to master every champion and lead your team to victory. The League of Legends universe is expanding into other game genres and Netflix shows, so now’s the time to get caught up.  League of Legendsreview It Takes Two In many ways, marriage is the ultimate multiplayer game. It Takes Two is a cooperative adventure that tasks two people with controlling a couple as they complete wacky challenges to repair their strained relationship. You’ll never know true love until you and your partner escape a giant cuckoo clock together. The King of Fighters XV The King of Fighters XV4.0 Excellent For finely tuned 2D fighting, look no further than The King of Fighters XV. Building off intriguing ideas introduced in previous entries, KOF XV gives you a massive character roster and an expressive, creative fighting system. Tournament features, multiplayer party modes, and rollback netcode make this one of the series' best entries. Mario Kart 8 Deluxe Mario Kart 8 Deluxe4.5 Excellent Mario Kart 8 Deluxe is absolutely everything you could want from Nintendo’s hugely popular kart racing series. It features gorgeous visuals, inventive tracks, and a revamped battle mode. In fact, Nintendo is still selling new courses, years after the game's 2017 debut. Mario Kart 8 Deluxe is so spectacular not even the blue shell can stop it.  Minecraft Minecraft4.5 Excellent Minecraft gives young people an unparalleled sense of freedom as they explore and build worlds, brick by brick. In fact, multiple builders can join the same game world for cooperative mining and crafting. Take it a step further by setting up your own Minecraft server, so you and friends can construct a private paradise.  Monster Hunter Rise Monster Hunter Rise4.5 Excellent Monster Hunter Rise finally makes hardened haters see the glory of Capcom’s monster-slaying series. You can craft new weapons and armors by defeating a monster menagerie inspired by Japanese mythology or swing through the air with new wirebug techniques. No beasts will stand in your way as you go beast hunting with friends. Mortal Kombat 1 Mortal Kombat 14.5 Excellent With Mortal Kombat 1, the famously bloody fighting game finally breaks free of its infamously stiff gameplay mechanics. Finishing foes has never been more fun thanks to high-flying air combos and custom tag-team Kameo fighter attacks. This reboot is as entertaining to play with friends as it is to watch with horrified onlookers. Mortal Kombat 1review Rocket League Rocket League4.5 Excellent “Cars playing soccer” is such a beautiful premise for an arcade sports game, and Rocket League perfectly pulls it off. Sure, you can just put the pedal to the metal and bash into the ball, hoping it goes into the goal. But the high-flying physics system creates enough depth for sensational tests of skill. The free-to-play season structure means you’ll always have a reason to return.  Rocket Leaguereview Splatoon 3 Splatoon 34.0 Excellent Only Nintendo could take the well-worn shooter genre and turn it into a game about punky squid kids squirting ink at each other. By making battles more about covering turf than blasting opponents, Splatoon 3 is a friendlier and more accessible shooter. However, you’ll need to stay on your toes with so many unique weapons and traversal options. Splatoon 3review StarCraft II StarCraft II: Legacy of the Void5.0 Outstanding StarCraft II is the best strategy game since chess. Whether you play as Terran, Zerg, or Protoss armies, you have access to perfectly balanced units for overcoming any opponent during real-time clashes. The StarCraft II trilogy even introduces free, cooperate multiplayer modes, so veterans can introduce newcomers to the fight. Street Fighter 6 Street Fighter 65.0 Outstanding Street Fighter 6 is worthy of its iconic name. With its bold new style, expansive new modes, exciting new roster, and competitive new gameplay systems, Street Fighter battles are more hype than ever. It's a multiplayer gaming masterpiece. Street Fighter 6review Streets of Rage 4 Streets of Rage 44.0 Excellent Streets of Rage 4 breathes new life into the aging beat ‘em up genre thanks to complex combat and stunning illustrated graphics. If smacking goons in solo fashion gets boring, team up with a friend for chaotic co-op action. You can even unlock retro versions of classic characters. Super Smash Bros. Ultimate Super Smash Bros. Ultimate4.5 Excellent Super Smash Bros. Ultimate combines countless characters, stages, modes, and music tracks to create the most incredible video game crossover of all time. It’s also a phenomenal platform-fighting game, speeding up the addictive combat and rebalancing advanced techniques. No matter how seriously you take it, no video game can scratch that satisfying multiplayer itch like Super Smash Bros.  Tekken 8 Tekken 8Tekken 8 is the latest and greatest entry in the venerable 3D fighting game franchise. Prove your worth at the King of Iron Fist Tournament by nimbly sidestepping, laying down painful combos, and activating new Heat Smash attacks. For the true Tekken multiplayer experience, fight your dad and throw him down a volcano. Tetris Effect: Connected Tetris Effect: Connected4.5 Excellent Tetris Effect: Connected makes the perfect puzzle game even better. Alongside traditional competitive Tetris multiplayer modes, Effect lets you team up for cooperative “Connected” journeys where you and your partners clear lines on the same massive board. Combine that with trance-inducing audiovisual stimuli, and you’ll never look at blocks the same way again.  #best #multiplayer #video #games
    ME.PCMAG.COM
    The Best Multiplayer Video Games for 2025
    Sometimes you want to play alone, whether your game of choice is a relaxing solitaire session or an engrossing, cinematic campaign. We get that. Still, some of the best gaming-related experiences come from moments shared with other people. After all, an excellent multiplayer mode makes a video game endlessly replayable and enables good times with local friends or strangers across the country—as long as the servers stay up. Our list of the best multiplayer games casts a wide net that includes console and PC games, competitive and cooperative titles, casual board games and serious esports fare, and, of course, battle royales, shooters, and fighters. If you're interested in playing a game with at least one other person, you'll find something that catches your eye here. These are the best multiplayer video games that you and your friends should play right now. Apex Legends From the ashes of Titanfall rose the best battle royale game. Respawn’s Apex Legends combines unbelievably fluid movement with impeccable gunplay and innovative team communication features. Each character’s unique abilities open strategic options on the expansive battlefield. Among Us Among Us (for iOS) 4.0 Excellent Among Us is more of a social experiment than a game. You and your friends play as crewmates attempting to repair a spaceship, but some players are deadly impostors who are picking off others. Constant lying and manipulation turn even the friendliest relationships into pure paranoia.  Clubhouse Games Clubhouse Games: 51 Worldwide Classics (for Nintendo Switch) 4.0 Excellent This compilation contains more than 50 classic games that have stood the test of time, including bowling, backgammon, and billiards. You can have fun with friends or family, but beware getting so heated that you'll never want to speak with them again. Online play ensures the good (and frustrating) times aren't limited to your immediate vicinity. Counter-Strike 2 Counter-Strike: Global Offensive 4.0 Excellent Counter-Strike is a founding father of multiplayer online shooters, and Counter-Strike 2 continues the legacy. In this long-awaited update to Counter-Strike: Global Offensive, you’ll find a massive community always ready to hop into Terrorist versus Counter-Terrorist tactical team action. Plus, it’s free! Death Stranding Death Stranding: Director's Cut (for PlayStation 5) 4.0 Excellent Hideo Kojima’s freaky odyssey about time rain and babies in jars is also one of the most fascinating and unconventional multiplayer games in recent memory. As you traverse the harsh wasteland, you can leave behind useful items, such as ladders and reports, that other players can use in their sessions. Destiny 2 Destiny 2 (for PlayStation 4) Destiny 2 is the looter-shooter that gives other looter-shooters envy. You gather the shiniest guns, the sickest armor, and show off your gear in front of fellow Guardians. Party-up and shoot your way through alien enemies and strongholds with Bungie’s perfect first-person shooter controls. The first taste is free, and regular, new content releases give you many reasons to keep gunning. Destiny 2 (for PlayStation 4) review Diablo IV 4.0 Excellent No action-RPG out-Diablos Diablo IV, a title that expands the familiar loot-grinding mechanics with massive, demon-filled zones. In terms of character builds, Diablo IV review Dota 2 Dota 2 (for PC) 4.5 Excellent What began as a mere mod has since become one of the most popular esports in the world. Dota 2 sets the standard for the MOBA genre, that strange hybrid between real-time strategy and team sports. New heroes give players constantly changing choices to consider. If you put in the effort to get really good at this game, the sky's the limit.  Fatal Fury: City of the Wolves Fatal Fury: City of the Wolves (for PC) 4.0 Excellent The King of Fighters series is great, but Fatal Fury: City of the Wolves resurrects the SNK fighter that started it all. Familiar faces like Terry Bogard and Mai Shiranui battle real-life guest characters like DJ Salavatore Gannaci and soccer star Cristiano Ronaldo in this excellent take on fundamental, 2D fisticuffs. Rich mechanics add depth to both offensive and defensive play, while comic book-inspired graphics give brawls a distinct visual identity. Crossplay multiplayer shines with rollback netcode. Fatal Fury: City of the Wolves (for PC) review Fortnite Fortnite (for PC) 3.5 Good Do you really need someone else to tell you about Fortnite? Originally a free battle royale mode for a failed multiplayer game, Fortnite became an absolute phenomenon. Every day, millions of children leap from the in-game battle bus to shoot each other and build elaborate structures, while dressed as their favorite brands. You can also hang out and watch concerts or (for some reason) documentaries on social issues.   Forza Horizon 5 Forza Horizon 5 (for PC) 4.5 Excellent Forza Horizon 5 appeals equally to serious automobile racing enthusiasts and anyone who just wants to drive aimlessly through beautifully rendered Mexican landscapes. Although largely similar to past entries, the new EventLab lets you create clever, custom courses to enjoy with friends. Halo Infinite Halo Infinite (for PC) 4.5 Excellent Halo single-handedly saved the Xbox, and proved that multiplayer shooters could thrive on home consoles. Halo Infinite doesn’t just reinvent the single-player campaign, it continues Halo’s history of excellent multiplayer modes, from capture the flag to random weapon fiestas. Plus, you can play for free, so finish the fight. Halo Infinite (for PC) review Jackbox Party Pack The Jackbox Party Pack 8 (for PC) The annual Jackbox Party Pack games consistently deliver the most hilarious social multiplayer experiences you’ll ever play. Design wacky t-shirts, come up with witty quips, and try to figure out which friend is faking it. Anyone can play, as long as they have a phone. With unique streaming features, even your audience can join the party. League of Legends League of Legends (for PC) 4.5 Excellent Free from any previous mod baggage, League of Legends is arguably the more accessible game when it comes to the MOBA heavy hitters. Still, it takes skill to master every champion and lead your team to victory. The League of Legends universe is expanding into other game genres and Netflix shows, so now’s the time to get caught up.  League of Legends (for PC) review It Takes Two In many ways, marriage is the ultimate multiplayer game. It Takes Two is a cooperative adventure that tasks two people with controlling a couple as they complete wacky challenges to repair their strained relationship. You’ll never know true love until you and your partner escape a giant cuckoo clock together. The King of Fighters XV The King of Fighters XV (for PC) 4.0 Excellent For finely tuned 2D fighting, look no further than The King of Fighters XV. Building off intriguing ideas introduced in previous entries, KOF XV gives you a massive character roster and an expressive, creative fighting system. Tournament features, multiplayer party modes, and rollback netcode make this one of the series' best entries. Mario Kart 8 Deluxe Mario Kart 8 Deluxe (for Nintendo Switch) 4.5 Excellent Mario Kart 8 Deluxe is absolutely everything you could want from Nintendo’s hugely popular kart racing series. It features gorgeous visuals, inventive tracks, and a revamped battle mode. In fact, Nintendo is still selling new courses, years after the game's 2017 debut. Mario Kart 8 Deluxe is so spectacular not even the blue shell can stop it.  Minecraft Minecraft (for PC) 4.5 Excellent Minecraft gives young people an unparalleled sense of freedom as they explore and build worlds, brick by brick. In fact, multiple builders can join the same game world for cooperative mining and crafting. Take it a step further by setting up your own Minecraft server, so you and friends can construct a private paradise.  Monster Hunter Rise Monster Hunter Rise (for Nintendo Switch) 4.5 Excellent Monster Hunter Rise finally makes hardened haters see the glory of Capcom’s monster-slaying series. You can craft new weapons and armors by defeating a monster menagerie inspired by Japanese mythology or swing through the air with new wirebug techniques. No beasts will stand in your way as you go beast hunting with friends. Mortal Kombat 1 Mortal Kombat 1 (For PC) 4.5 Excellent With Mortal Kombat 1, the famously bloody fighting game finally breaks free of its infamously stiff gameplay mechanics. Finishing foes has never been more fun thanks to high-flying air combos and custom tag-team Kameo fighter attacks. This reboot is as entertaining to play with friends as it is to watch with horrified onlookers. Mortal Kombat 1 (For PC) review Rocket League Rocket League (for PC) 4.5 Excellent “Cars playing soccer” is such a beautiful premise for an arcade sports game, and Rocket League perfectly pulls it off. Sure, you can just put the pedal to the metal and bash into the ball, hoping it goes into the goal. But the high-flying physics system creates enough depth for sensational tests of skill. The free-to-play season structure means you’ll always have a reason to return.  Rocket League (for PC) review Splatoon 3 Splatoon 3 (for Nintendo Switch) 4.0 Excellent Only Nintendo could take the well-worn shooter genre and turn it into a game about punky squid kids squirting ink at each other. By making battles more about covering turf than blasting opponents, Splatoon 3 is a friendlier and more accessible shooter. However, you’ll need to stay on your toes with so many unique weapons and traversal options. Splatoon 3 (for Nintendo Switch) review StarCraft II StarCraft II: Legacy of the Void (for PC) 5.0 Outstanding StarCraft II is the best strategy game since chess. Whether you play as Terran, Zerg, or Protoss armies, you have access to perfectly balanced units for overcoming any opponent during real-time clashes. The StarCraft II trilogy even introduces free, cooperate multiplayer modes, so veterans can introduce newcomers to the fight. Street Fighter 6 Street Fighter 6 (for PC) 5.0 Outstanding Street Fighter 6 is worthy of its iconic name. With its bold new style (graffiti in motion!), expansive new modes (worldwide online Battle Hub!), exciting new roster (Kimberly!), and competitive new gameplay systems (Drive Gauge!), Street Fighter battles are more hype than ever. It's a multiplayer gaming masterpiece. Street Fighter 6 (for PC) review Streets of Rage 4 Streets of Rage 4 (for PC) 4.0 Excellent Streets of Rage 4 breathes new life into the aging beat ‘em up genre thanks to complex combat and stunning illustrated graphics. If smacking goons in solo fashion gets boring, team up with a friend for chaotic co-op action. You can even unlock retro versions of classic characters. Super Smash Bros. Ultimate Super Smash Bros. Ultimate (for Nintendo Switch) 4.5 Excellent Super Smash Bros. Ultimate combines countless characters, stages, modes, and music tracks to create the most incredible video game crossover of all time. It’s also a phenomenal platform-fighting game, speeding up the addictive combat and rebalancing advanced techniques. No matter how seriously you take it, no video game can scratch that satisfying multiplayer itch like Super Smash Bros.  Tekken 8 Tekken 8 (for PC) Tekken 8 is the latest and greatest entry in the venerable 3D fighting game franchise. Prove your worth at the King of Iron Fist Tournament by nimbly sidestepping, laying down painful combos, and activating new Heat Smash attacks. For the true Tekken multiplayer experience, fight your dad and throw him down a volcano. Tetris Effect: Connected Tetris Effect: Connected (for Xbox Series S) 4.5 Excellent Tetris Effect: Connected makes the perfect puzzle game even better. Alongside traditional competitive Tetris multiplayer modes, Effect lets you team up for cooperative “Connected” journeys where you and your partners clear lines on the same massive board. Combine that with trance-inducing audiovisual stimuli, and you’ll never look at blocks the same way again. 
    12 Comentários 0 Compartilhamentos
  • ⚡️ Instant Impostors – One-click impostor generation for Blender (Eevee)

    ⚡️ Instant Impostors – One-click impostor generation for BlenderOptimize smart. Create big. /...

    Need complex assets without the heavy geometry? Instant Impostors bakes 3D objects into lightweight billboard-style impostors—perfect for real-time scenes.

    Ideal for game dev, massive environments, archviz, and background optimization
    One-click setup – just select your object & generate an impostor
    Boosts viewport & render performance dramatically
    Works with Eevee, includes LODs, supports lighting/shadow baking
    Royalty-free license – use in personal and commercial projects

    #b3d #blender3d #blender #BlenderAddon #GameDev #Optimization #Eevee #ImpostorSystem #tree
    #instant #impostors #oneclick #impostor #generation
    ⚡️ Instant Impostors – One-click impostor generation for Blender (Eevee)
    ⚡️ Instant Impostors – One-click impostor generation for BlenderOptimize smart. Create big. 👉 /... Need complex assets without the heavy geometry? Instant Impostors bakes 3D objects into lightweight billboard-style impostors—perfect for real-time scenes. 🎮 Ideal for game dev, massive environments, archviz, and background optimization 🖱️ One-click setup – just select your object & generate an impostor 🚀 Boosts viewport & render performance dramatically 🌗 Works with Eevee, includes LODs, supports lighting/shadow baking 📜 Royalty-free license – use in personal and commercial projects #b3d #blender3d #blender #BlenderAddon #GameDev #Optimization #Eevee #ImpostorSystem #tree #instant #impostors #oneclick #impostor #generation
    WWW.YOUTUBE.COM
    ⚡️ Instant Impostors – One-click impostor generation for Blender (Eevee)
    ⚡️ Instant Impostors – One-click impostor generation for Blender (Eevee) Optimize smart. Create big. 👉 https://superhivemarket.com/products/... Need complex assets without the heavy geometry? Instant Impostors bakes 3D objects into lightweight billboard-style impostors—perfect for real-time scenes. 🎮 Ideal for game dev, massive environments, archviz, and background optimization 🖱️ One-click setup – just select your object & generate an impostor 🚀 Boosts viewport & render performance dramatically 🌗 Works with Eevee, includes LODs, supports lighting/shadow baking 📜 Royalty-free license – use in personal and commercial projects #b3d #blender3d #blender #BlenderAddon #GameDev #Optimization #Eevee #ImpostorSystem #tree (Feed generated with FetchRSS)
    0 Comentários 0 Compartilhamentos
  • Magentic-UI, an experimental human-centered web agent

    Modern productivity is rooted in the web—from searching for information and filling in forms to navigating dashboards. Yet, many of these tasks remain manual and repetitive. Today, we are introducing Magentic-UI, a new open-source research prototype of a human-centered agent that is meant to help researchers study open questions on human-in-the-loop approaches and oversight mechanisms for AI agents. This prototype collaborates with users on web-based tasks and operates in real time over a web browser. Unlike other computer use agents that aim for full autonomy, Magentic-UI offers a transparent and controllable experience for tasks that are action-oriented and
    Magentic-UI builds on Magentic-One, a powerful multi-agent team we released last year, and is powered by AutoGen, our leading agent framework. It is available under MIT license atand on Azure AI Foundry Labs, the hub where developers, startups, and enterprises can explore groundbreaking innovations from Microsoft Research. Magentic-UI is integrated with Azure AI Foundry models and agents. Learn more about how to integrate Azure AI agents into the Magentic-UI multi-agent architecture by following this code sample. 
    Magentic-UI can perform tasks that require browsing the web, writing and executing Python and shell code, and understanding files. Its key features include:

    Collaborative planning with users. Magentic-UI allows users to directly modify its plan through a plan editor or by providing textual feedback before Magentic-UI executes any actions. 
    Collaborative execution with users. Users can pause the system and give feedback in natural language or demonstrate it by directly taking control of the browser.
    Safety with human-in-the-loop. Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors.
    Safety with human-in-the-loop. Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors. 
    Learning from experience. Magentic-UI can learn and save plans from previous interactions to improve task completion for future tasks. 

    Figure 1: Screenshot of Magentic-UI actively performing a task. The left side of the screen shows Magentic-UI stating its plan and progress to accomplish a user’s complex goal. The right side shows the browser Magentic-UI is controlling. 
    How is Magentic-UI human-centered?
    While many web agents promise full autonomy, in practice users can be left unsure of what the agent can do, what it is currently doing, and whether they have enough control to intervene when something goes wrong or doesn’t occur as expected. By contrast, Magentic-UI considers user needs at every stage of interaction. We followed a human-centered design methodology in building Magentic-UI by prototyping and obtaining feedback from pilot users during its design. 
    Figure 2: Co-planning – Users can collaboratively plan with Magentic-UI.
    For example, after a person specifies and before Magentic-UI even begins to execute, it creates a clear step-by-step plan that outlines what it would do to accomplish the task. People can collaborate with Magentic-UI to modify this plan and then give final approval for Magentic-UI to begin execution. This is crucial as users may have expectations of how the task should be completed; communicating that information could significantly improve agent performance. We call this feature co-planning.
    During execution, Magentic-UI shows in real time what specific actions it’s about to take. For example, whether it is about to click on a button or input a search query. It also shows in real time what it observed on the web pages it is visiting. Users can take control of the action at any point in time and give control back to the agent. We call this feature co-tasking.
    Figure 3: Co-tasking – Magentic-UI provides real-time updates about what it is about to do and what it already did, allowing users to collaboratively complete tasks with the agent.
    Figure 4: Action-guards – Magentic-UI will ask users for permission before executing actions that it deems consequential or important. 
    Additionally, Magentic-UI asks for user permission before performing actions that are deemed irreversible, such as closing a tab or clicking a button with side effects. We call these “action guards”. The user can also configure Magentic-UI’s action guards to always ask for permission before performing any action. If the user deems an action risky, they can reject it. 

    Figure 5: Plan learning – Once a task is successfully completed, users can request Magentic-UI to learn a step-by-step plan from this experience.
    After execution, the user can ask Magentic-UI to reflect on the conversation and infer and save a step-by-step plan for future similar tasks. Users can view and modify saved plans for Magentic-UI to reuse in the future in a saved-plans gallery. In a future session, users can launch Magentic-UI with the saved plan to either execute the same task again, like checking the price of a specific flight, or use the plan as a guide to help complete similar tasks, such as checking the price of a different type of flight. 
    Combined, these four features—co-planning, co-tasking, action guards, and plan learning—enable users to collaborate effectively with Magentic-UI.
    Architecture
    Magentic-UI’s underlying system is a team of specialized agents adapted from AutoGen’s Magentic-One system. The agents work together to create a modular system:

    Orchestrator is the lead agent, powered by a large language model, that performs co-planning with the user, decides when to ask the user for feedback, and delegates sub-tasks to the remaining agents to complete.
    WebSurfer is an LLM agent equipped with a web browser that it can control. Given a request by the Orchestrator, it can click, type, scroll, and visit pages in multiple rounds to complete the request from the Orchestrator.
    Coder is an LLM agent equipped with a Docker code-execution container. It can write and execute Python and shell commands and provide a response back to the Orchestrator.
    FileSurfer is an LLM agent equipped with a Docker code-execution container and file-conversion tools from the MarkItDownpackage. It can locate files in the directory controlled by Magentic-UI, convert files to markdown, and answer questions about them.

    Figure 6: System architecture diagram of Magentic-UI
    To interact with Magentic-UI, users can enter a text message and attach images. In response, Magentic-UI creates a natural-language step-by-step plan with which users can interact through a plan-editing interface. Users can add, delete, edit, regenerate steps, and write follow-up messages to iterate on the plan. While the user editing the plan adds an upfront cost to the interaction, it can potentially save a significant amount of time in the agent executing the plan and increase its chance at success.
    The plan is stored inside the Orchestrator and is used to execute the task. For each step of the plan, the Orchestrator determines which of the agentsor the user should complete the step. Once that decision is made, the Orchestrator sends a request to one of the agents or the user and waits for a response. After the response is received, the Orchestrator decides whether that step is complete. If it is, the Orchestrator moves on to the following step.
    Once all steps are completed, the Orchestrator generates a final answer that is presented to the user. If, while executing any of the steps, the Orchestrator decides that the plan is inadequate, the Orchestrator can replan with user permission and start executing a new plan.
    All intermediate progress steps are clearly displayed to the user. Furthermore, the user can pause the execution of the plan and send additional requests or feedback. The user can also configure through the interface whether agent actionsrequire approval.
    Evaluating Magentic-UI
    Magentic-UI innovates through its ability to integrate human feedback in its planning and execution of tasks. We performed a preliminary automated evaluation to showcase this ability on the GAIA benchmarkfor agents with a user-simulation experiment.
    Evaluation with simulated users
    Figure 7: Comparison on the GAIA validation set of the accuracy of Magentic-One, Magentic-UI in autonomous mode, Magentic-UI with a simulated user powered by a smarter LLM than the MAGUI agents, Magentic-UI with a simulated user that has a\access to side information about the tasks, and human performance. This shows that human-in-the-loop can improve the accuracy of autonomous agents, bridging the gap to human performance at a fraction of the cost.
    GAIA is a benchmark for general AI assistants, with multimodal question-answer pairs that are challenging, requiring the agents to navigate the web, process files, and execute code. The traditional evaluation setup with GAIA assumes the system will autonomously complete the task and return an answer, which is compared to the ground-truth answer. 
    To evaluate the human-in-the-loop capabilities of Magentic-UI, we transform GAIA into an interactive benchmark by introducing the concept of a simulated user. Simulated users provide value in two ways: by having specific expertise that the agent may not possess, and by providing guidance on how the task should be performed.
    We experiment with two types of simulated users to show the value of human-in-the-loop:a simulated user that is more intelligent than the Magentic-UI agents anda simulated user with the same intelligence as Magentic-UI agents but with additional information about the task. During co-planning, Magentic-UI takes feedback from this simulated user to improve its plan. During co-tasking, Magentic-UI can ask theuser for help when it gets stuck. Finally, if Magentic-UI does not provide a final answer, then the simulated user provides an answer instead.
    The simulated user is an LLM without any tools, instructed to interact with Magentic-UI the way we expect a human would act. The first type of simulated user relies on OpenAI’s o4-mini, more performant at many tasks than the one powering the Magentic-UI agents. For the second type of simulated user, we use GPT-4o for both the simulated user and the rest of the agents, but the user has access to side information about each task. Each task in GAIA has side information, which includes a human-written plan to solve the task. While this plan is not used as input in the traditional benchmark, in our interactive setting we provide this information to the second type of simulated user,which is powered by an LLM so that it can mimic a knowledgeable user. Importantly, we tuned our simulated user so as not to reveal the ground-truth answer directly as the answer is usually found inside the human written plan. Instead, it is prompted to guide Magentic-UI indirectly. We found that this tuning prevented the simulated user from inadvertently revealing the answer in all but 6% of tasks when Magentic-UI provides a final answer. 
    On the validation subset of GAIA, we show the results of Magentic-One operating in autonomous mode, Magentic-UI operating in autonomous mode, Magentic-UI with simulated user, Magentic-UI with simulated user, and human performance. We first note that Magentic-UI in autonomous mode is within a margin of error of the performance of Magentic-One. Note that the same LLMis used for Magentic-UI and Magentic-One.
    Magentic-UI with the simulated user that has access to side information improves the accuracy of autonomous Magentic-UI by 71%, from a 30.3% task-completion rate to a 51.9% task-completion rate. Moreover, Magentic-UI only asks for help from the simulated user in 10% of tasks and relies on the simulated user for the final answer in 18% of tasks. And in those tasks where it does ask for help, it asks for help on average 1.1 times. Magentic-UI with the simulated user powered by a smarter model improves to 42.6% where Magentic-UI asks for help in only 4.3% of tasks, asking for help an average of 1.7 times in those tasks. This demonstrates the potential of even lightweight human feedback for improving performanceof autonomous agents, especially at a fraction of the cost compared to people completing tasks entirely manually. 
    Learning and reusing plans
    As described above, once Magentic-UI completes a task, users have the option for Magentic-UI to learn a plan based on the execution of the task. These plans are saved in a plan gallery, which users and Magentic-UI can access in the future.
    The user can select a plan from the plan gallery, which is displayed by clicking on the Saved Plans button. Alternatively, as a user enters a task that closely matches a previous task, the saved plan will be displayed even before the user is done typing. If no identical task is found, Magentic-UI can use AutoGen’s Task-Centric Memoryto retrieve plans for any similar tasks. Our preliminary evaluations show that this retrieval is highly accurate, and when recalling a saved plan can be around 3x faster than generating a new plan. Once a plan is recalled or generated, the user can always accept it, modify it, or ask Magentic-UI to modify it for the specific task at hand. 
    Safety and control
    Magentic-UI can surf the live internet and execute code. With such capabilities, we need to ensure that Magentic-UI acts in a safe and secure manner. The following features, design decisions, and evaluations were made to ensure this:

    Allow-list: Users can set a list of websites that Magentic-UI is allowed to access. If Magentic-UI needs to access a website outside of the allow-list, users must explicitly approve it through the interface
    Anytime interruptions: At any point of Magentic-UI completing the task, the user can interrupt Magentic-UI and stop any pending code execution or web browsing.
    Docker sandboxing: Magentic-UI controls a browser that is launched inside a Docker container with no credentials, which avoids risks with logged-in accounts and credentials. Moreover, any code execution is also performed inside a separate Docker container to avoid affecting the host environment in which Magentic-UI is running. This is illustrated in the system architecture of Magentic-UI.
    Detection and approval of irreversible agent actions: Users can configure an action-approval policyto determine which actions Magentic-UI can perform without user approval. In the extreme, users can specify that any actionneeds explicit user approval. Users must press an “Accept” or “Deny” button for each action.

    In addition to the above design decisions, we performed a red-team evaluation of Magentic-UI on a set of internal scenarios, which we developed to challenge the security and safety of Magentic-UI. Such scenarios include cross-site prompt injection attacks, where web pages contain malicious instructions distinct from the user’s original intent. It also contains scenarios comparable to phishing, which try to trick Magentic-UI into entering sensitive information, or granting permissions on impostor sites. In our preliminary evaluations, we found that Magentic-UI either refuses to complete the requests, stops to ask the user, or, as a final safety measure, is eventually unable to complete the request due to Docker sandboxing. We have found that this layered approach is effective for thwarting these attacks.
    We have also released transparency notes, which can be found at:Open research questions 
    Magentic-UI provides a tool for researchers to study critical questions in agentic systems and particularly on human-agent interaction. In a previous report, we outlined 12 questions for human-agent communication, and Magentic-UI provides a vehicle to study these questions in a realistic setting. A key question among these is how we enable humans to efficiently intervene and provide feedback to the agent while executing a task. Humans should not have to constantly watch the agent. Ideally, the agent should know when to reach out for help and provide the necessary context for the human to assist it. A second question is about safety. As agents interact with the live web, they may become prone to attacks from malicious actors. We need to study what necessary safeguards are needed to protect the human from side effects without adding a heavy burden on the human to verify every agent action. There are also many other questions surrounding security, personalization, and learning that Magentic-UI can help with studying. 
    Conclusion
    Magentic-UI is an open-source agent prototype that works with people to complete complex tasks that require multi-step planning and browser use. As agentic systems expand in the scope of tasks they can complete, Magentic-UI’s design enables better transparency into agent actions and enables human control to ensure safety and reliability. Moreover, by facilitating human intervention, we can improve performance while still reducing human cost in completing tasks on aggregate. Today we have released the first version of Magentic-UI. Looking ahead, we plan to continue developing it in the open with the goal of improving its capabilities and answering research questions on human-agent collaboration. We invite the research community to extend and reuse Magentic-UI for their scientific explorations and domains. 
    Opens in a new tab
    #magenticui #experimental #humancentered #web #agent
    Magentic-UI, an experimental human-centered web agent
    Modern productivity is rooted in the web—from searching for information and filling in forms to navigating dashboards. Yet, many of these tasks remain manual and repetitive. Today, we are introducing Magentic-UI, a new open-source research prototype of a human-centered agent that is meant to help researchers study open questions on human-in-the-loop approaches and oversight mechanisms for AI agents. This prototype collaborates with users on web-based tasks and operates in real time over a web browser. Unlike other computer use agents that aim for full autonomy, Magentic-UI offers a transparent and controllable experience for tasks that are action-oriented and Magentic-UI builds on Magentic-One, a powerful multi-agent team we released last year, and is powered by AutoGen, our leading agent framework. It is available under MIT license atand on Azure AI Foundry Labs, the hub where developers, startups, and enterprises can explore groundbreaking innovations from Microsoft Research. Magentic-UI is integrated with Azure AI Foundry models and agents. Learn more about how to integrate Azure AI agents into the Magentic-UI multi-agent architecture by following this code sample.  Magentic-UI can perform tasks that require browsing the web, writing and executing Python and shell code, and understanding files. Its key features include: Collaborative planning with users. Magentic-UI allows users to directly modify its plan through a plan editor or by providing textual feedback before Magentic-UI executes any actions.  Collaborative execution with users. Users can pause the system and give feedback in natural language or demonstrate it by directly taking control of the browser. Safety with human-in-the-loop. Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors. Safety with human-in-the-loop. Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors.  Learning from experience. Magentic-UI can learn and save plans from previous interactions to improve task completion for future tasks.  Figure 1: Screenshot of Magentic-UI actively performing a task. The left side of the screen shows Magentic-UI stating its plan and progress to accomplish a user’s complex goal. The right side shows the browser Magentic-UI is controlling.  How is Magentic-UI human-centered? While many web agents promise full autonomy, in practice users can be left unsure of what the agent can do, what it is currently doing, and whether they have enough control to intervene when something goes wrong or doesn’t occur as expected. By contrast, Magentic-UI considers user needs at every stage of interaction. We followed a human-centered design methodology in building Magentic-UI by prototyping and obtaining feedback from pilot users during its design.  Figure 2: Co-planning – Users can collaboratively plan with Magentic-UI. For example, after a person specifies and before Magentic-UI even begins to execute, it creates a clear step-by-step plan that outlines what it would do to accomplish the task. People can collaborate with Magentic-UI to modify this plan and then give final approval for Magentic-UI to begin execution. This is crucial as users may have expectations of how the task should be completed; communicating that information could significantly improve agent performance. We call this feature co-planning. During execution, Magentic-UI shows in real time what specific actions it’s about to take. For example, whether it is about to click on a button or input a search query. It also shows in real time what it observed on the web pages it is visiting. Users can take control of the action at any point in time and give control back to the agent. We call this feature co-tasking. Figure 3: Co-tasking – Magentic-UI provides real-time updates about what it is about to do and what it already did, allowing users to collaboratively complete tasks with the agent. Figure 4: Action-guards – Magentic-UI will ask users for permission before executing actions that it deems consequential or important.  Additionally, Magentic-UI asks for user permission before performing actions that are deemed irreversible, such as closing a tab or clicking a button with side effects. We call these “action guards”. The user can also configure Magentic-UI’s action guards to always ask for permission before performing any action. If the user deems an action risky, they can reject it.  Figure 5: Plan learning – Once a task is successfully completed, users can request Magentic-UI to learn a step-by-step plan from this experience. After execution, the user can ask Magentic-UI to reflect on the conversation and infer and save a step-by-step plan for future similar tasks. Users can view and modify saved plans for Magentic-UI to reuse in the future in a saved-plans gallery. In a future session, users can launch Magentic-UI with the saved plan to either execute the same task again, like checking the price of a specific flight, or use the plan as a guide to help complete similar tasks, such as checking the price of a different type of flight.  Combined, these four features—co-planning, co-tasking, action guards, and plan learning—enable users to collaborate effectively with Magentic-UI. Architecture Magentic-UI’s underlying system is a team of specialized agents adapted from AutoGen’s Magentic-One system. The agents work together to create a modular system: Orchestrator is the lead agent, powered by a large language model, that performs co-planning with the user, decides when to ask the user for feedback, and delegates sub-tasks to the remaining agents to complete. WebSurfer is an LLM agent equipped with a web browser that it can control. Given a request by the Orchestrator, it can click, type, scroll, and visit pages in multiple rounds to complete the request from the Orchestrator. Coder is an LLM agent equipped with a Docker code-execution container. It can write and execute Python and shell commands and provide a response back to the Orchestrator. FileSurfer is an LLM agent equipped with a Docker code-execution container and file-conversion tools from the MarkItDownpackage. It can locate files in the directory controlled by Magentic-UI, convert files to markdown, and answer questions about them. Figure 6: System architecture diagram of Magentic-UI To interact with Magentic-UI, users can enter a text message and attach images. In response, Magentic-UI creates a natural-language step-by-step plan with which users can interact through a plan-editing interface. Users can add, delete, edit, regenerate steps, and write follow-up messages to iterate on the plan. While the user editing the plan adds an upfront cost to the interaction, it can potentially save a significant amount of time in the agent executing the plan and increase its chance at success. The plan is stored inside the Orchestrator and is used to execute the task. For each step of the plan, the Orchestrator determines which of the agentsor the user should complete the step. Once that decision is made, the Orchestrator sends a request to one of the agents or the user and waits for a response. After the response is received, the Orchestrator decides whether that step is complete. If it is, the Orchestrator moves on to the following step. Once all steps are completed, the Orchestrator generates a final answer that is presented to the user. If, while executing any of the steps, the Orchestrator decides that the plan is inadequate, the Orchestrator can replan with user permission and start executing a new plan. All intermediate progress steps are clearly displayed to the user. Furthermore, the user can pause the execution of the plan and send additional requests or feedback. The user can also configure through the interface whether agent actionsrequire approval. Evaluating Magentic-UI Magentic-UI innovates through its ability to integrate human feedback in its planning and execution of tasks. We performed a preliminary automated evaluation to showcase this ability on the GAIA benchmarkfor agents with a user-simulation experiment. Evaluation with simulated users Figure 7: Comparison on the GAIA validation set of the accuracy of Magentic-One, Magentic-UI in autonomous mode, Magentic-UI with a simulated user powered by a smarter LLM than the MAGUI agents, Magentic-UI with a simulated user that has a\access to side information about the tasks, and human performance. This shows that human-in-the-loop can improve the accuracy of autonomous agents, bridging the gap to human performance at a fraction of the cost. GAIA is a benchmark for general AI assistants, with multimodal question-answer pairs that are challenging, requiring the agents to navigate the web, process files, and execute code. The traditional evaluation setup with GAIA assumes the system will autonomously complete the task and return an answer, which is compared to the ground-truth answer.  To evaluate the human-in-the-loop capabilities of Magentic-UI, we transform GAIA into an interactive benchmark by introducing the concept of a simulated user. Simulated users provide value in two ways: by having specific expertise that the agent may not possess, and by providing guidance on how the task should be performed. We experiment with two types of simulated users to show the value of human-in-the-loop:a simulated user that is more intelligent than the Magentic-UI agents anda simulated user with the same intelligence as Magentic-UI agents but with additional information about the task. During co-planning, Magentic-UI takes feedback from this simulated user to improve its plan. During co-tasking, Magentic-UI can ask theuser for help when it gets stuck. Finally, if Magentic-UI does not provide a final answer, then the simulated user provides an answer instead. The simulated user is an LLM without any tools, instructed to interact with Magentic-UI the way we expect a human would act. The first type of simulated user relies on OpenAI’s o4-mini, more performant at many tasks than the one powering the Magentic-UI agents. For the second type of simulated user, we use GPT-4o for both the simulated user and the rest of the agents, but the user has access to side information about each task. Each task in GAIA has side information, which includes a human-written plan to solve the task. While this plan is not used as input in the traditional benchmark, in our interactive setting we provide this information to the second type of simulated user,which is powered by an LLM so that it can mimic a knowledgeable user. Importantly, we tuned our simulated user so as not to reveal the ground-truth answer directly as the answer is usually found inside the human written plan. Instead, it is prompted to guide Magentic-UI indirectly. We found that this tuning prevented the simulated user from inadvertently revealing the answer in all but 6% of tasks when Magentic-UI provides a final answer.  On the validation subset of GAIA, we show the results of Magentic-One operating in autonomous mode, Magentic-UI operating in autonomous mode, Magentic-UI with simulated user, Magentic-UI with simulated user, and human performance. We first note that Magentic-UI in autonomous mode is within a margin of error of the performance of Magentic-One. Note that the same LLMis used for Magentic-UI and Magentic-One. Magentic-UI with the simulated user that has access to side information improves the accuracy of autonomous Magentic-UI by 71%, from a 30.3% task-completion rate to a 51.9% task-completion rate. Moreover, Magentic-UI only asks for help from the simulated user in 10% of tasks and relies on the simulated user for the final answer in 18% of tasks. And in those tasks where it does ask for help, it asks for help on average 1.1 times. Magentic-UI with the simulated user powered by a smarter model improves to 42.6% where Magentic-UI asks for help in only 4.3% of tasks, asking for help an average of 1.7 times in those tasks. This demonstrates the potential of even lightweight human feedback for improving performanceof autonomous agents, especially at a fraction of the cost compared to people completing tasks entirely manually.  Learning and reusing plans As described above, once Magentic-UI completes a task, users have the option for Magentic-UI to learn a plan based on the execution of the task. These plans are saved in a plan gallery, which users and Magentic-UI can access in the future. The user can select a plan from the plan gallery, which is displayed by clicking on the Saved Plans button. Alternatively, as a user enters a task that closely matches a previous task, the saved plan will be displayed even before the user is done typing. If no identical task is found, Magentic-UI can use AutoGen’s Task-Centric Memoryto retrieve plans for any similar tasks. Our preliminary evaluations show that this retrieval is highly accurate, and when recalling a saved plan can be around 3x faster than generating a new plan. Once a plan is recalled or generated, the user can always accept it, modify it, or ask Magentic-UI to modify it for the specific task at hand.  Safety and control Magentic-UI can surf the live internet and execute code. With such capabilities, we need to ensure that Magentic-UI acts in a safe and secure manner. The following features, design decisions, and evaluations were made to ensure this: Allow-list: Users can set a list of websites that Magentic-UI is allowed to access. If Magentic-UI needs to access a website outside of the allow-list, users must explicitly approve it through the interface Anytime interruptions: At any point of Magentic-UI completing the task, the user can interrupt Magentic-UI and stop any pending code execution or web browsing. Docker sandboxing: Magentic-UI controls a browser that is launched inside a Docker container with no credentials, which avoids risks with logged-in accounts and credentials. Moreover, any code execution is also performed inside a separate Docker container to avoid affecting the host environment in which Magentic-UI is running. This is illustrated in the system architecture of Magentic-UI. Detection and approval of irreversible agent actions: Users can configure an action-approval policyto determine which actions Magentic-UI can perform without user approval. In the extreme, users can specify that any actionneeds explicit user approval. Users must press an “Accept” or “Deny” button for each action. In addition to the above design decisions, we performed a red-team evaluation of Magentic-UI on a set of internal scenarios, which we developed to challenge the security and safety of Magentic-UI. Such scenarios include cross-site prompt injection attacks, where web pages contain malicious instructions distinct from the user’s original intent. It also contains scenarios comparable to phishing, which try to trick Magentic-UI into entering sensitive information, or granting permissions on impostor sites. In our preliminary evaluations, we found that Magentic-UI either refuses to complete the requests, stops to ask the user, or, as a final safety measure, is eventually unable to complete the request due to Docker sandboxing. We have found that this layered approach is effective for thwarting these attacks. We have also released transparency notes, which can be found at:Open research questions  Magentic-UI provides a tool for researchers to study critical questions in agentic systems and particularly on human-agent interaction. In a previous report, we outlined 12 questions for human-agent communication, and Magentic-UI provides a vehicle to study these questions in a realistic setting. A key question among these is how we enable humans to efficiently intervene and provide feedback to the agent while executing a task. Humans should not have to constantly watch the agent. Ideally, the agent should know when to reach out for help and provide the necessary context for the human to assist it. A second question is about safety. As agents interact with the live web, they may become prone to attacks from malicious actors. We need to study what necessary safeguards are needed to protect the human from side effects without adding a heavy burden on the human to verify every agent action. There are also many other questions surrounding security, personalization, and learning that Magentic-UI can help with studying.  Conclusion Magentic-UI is an open-source agent prototype that works with people to complete complex tasks that require multi-step planning and browser use. As agentic systems expand in the scope of tasks they can complete, Magentic-UI’s design enables better transparency into agent actions and enables human control to ensure safety and reliability. Moreover, by facilitating human intervention, we can improve performance while still reducing human cost in completing tasks on aggregate. Today we have released the first version of Magentic-UI. Looking ahead, we plan to continue developing it in the open with the goal of improving its capabilities and answering research questions on human-agent collaboration. We invite the research community to extend and reuse Magentic-UI for their scientific explorations and domains.  Opens in a new tab #magenticui #experimental #humancentered #web #agent
    WWW.MICROSOFT.COM
    Magentic-UI, an experimental human-centered web agent
    Modern productivity is rooted in the web—from searching for information and filling in forms to navigating dashboards. Yet, many of these tasks remain manual and repetitive. Today, we are introducing Magentic-UI, a new open-source research prototype of a human-centered agent that is meant to help researchers study open questions on human-in-the-loop approaches and oversight mechanisms for AI agents. This prototype collaborates with users on web-based tasks and operates in real time over a web browser. Unlike other computer use agents that aim for full autonomy, Magentic-UI offers a transparent and controllable experience for tasks that are action-oriented and Magentic-UI builds on Magentic-One (opens in new tab), a powerful multi-agent team we released last year, and is powered by AutoGen (opens in new tab), our leading agent framework. It is available under MIT license at https://github.com/microsoft/Magentic-UI (opens in new tab) and on Azure AI Foundry Labs (opens in new tab), the hub where developers, startups, and enterprises can explore groundbreaking innovations from Microsoft Research. Magentic-UI is integrated with Azure AI Foundry models and agents. Learn more about how to integrate Azure AI agents into the Magentic-UI multi-agent architecture by following this code sample (opens in new tab).  Magentic-UI can perform tasks that require browsing the web, writing and executing Python and shell code, and understanding files. Its key features include: Collaborative planning with users (co-planning). Magentic-UI allows users to directly modify its plan through a plan editor or by providing textual feedback before Magentic-UI executes any actions.  Collaborative execution with users (co-tasking). Users can pause the system and give feedback in natural language or demonstrate it by directly taking control of the browser. Safety with human-in-the-loop (action guards). Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors. Safety with human-in-the-loop. Magentic-UI seeks user approval before executing potentially irreversible actions, and the user can specify how often Magentic-UI needs approvals. Furthermore, Magentic-UI is sandboxed for the safe operation of tools such as browsers and code executors.  Learning from experience (plan learning). Magentic-UI can learn and save plans from previous interactions to improve task completion for future tasks.  Figure 1: Screenshot of Magentic-UI actively performing a task. The left side of the screen shows Magentic-UI stating its plan and progress to accomplish a user’s complex goal. The right side shows the browser Magentic-UI is controlling.  How is Magentic-UI human-centered? While many web agents promise full autonomy, in practice users can be left unsure of what the agent can do, what it is currently doing, and whether they have enough control to intervene when something goes wrong or doesn’t occur as expected. By contrast, Magentic-UI considers user needs at every stage of interaction. We followed a human-centered design methodology in building Magentic-UI by prototyping and obtaining feedback from pilot users during its design.  Figure 2: Co-planning – Users can collaboratively plan with Magentic-UI. For example, after a person specifies and before Magentic-UI even begins to execute, it creates a clear step-by-step plan that outlines what it would do to accomplish the task. People can collaborate with Magentic-UI to modify this plan and then give final approval for Magentic-UI to begin execution. This is crucial as users may have expectations of how the task should be completed; communicating that information could significantly improve agent performance. We call this feature co-planning. During execution, Magentic-UI shows in real time what specific actions it’s about to take. For example, whether it is about to click on a button or input a search query. It also shows in real time what it observed on the web pages it is visiting. Users can take control of the action at any point in time and give control back to the agent. We call this feature co-tasking. Figure 3: Co-tasking – Magentic-UI provides real-time updates about what it is about to do and what it already did, allowing users to collaboratively complete tasks with the agent. Figure 4: Action-guards – Magentic-UI will ask users for permission before executing actions that it deems consequential or important.  Additionally, Magentic-UI asks for user permission before performing actions that are deemed irreversible, such as closing a tab or clicking a button with side effects. We call these “action guards”. The user can also configure Magentic-UI’s action guards to always ask for permission before performing any action. If the user deems an action risky (e.g., paying for an item), they can reject it.  Figure 5: Plan learning – Once a task is successfully completed, users can request Magentic-UI to learn a step-by-step plan from this experience. After execution, the user can ask Magentic-UI to reflect on the conversation and infer and save a step-by-step plan for future similar tasks. Users can view and modify saved plans for Magentic-UI to reuse in the future in a saved-plans gallery. In a future session, users can launch Magentic-UI with the saved plan to either execute the same task again, like checking the price of a specific flight, or use the plan as a guide to help complete similar tasks, such as checking the price of a different type of flight.  Combined, these four features—co-planning, co-tasking, action guards, and plan learning—enable users to collaborate effectively with Magentic-UI. Architecture Magentic-UI’s underlying system is a team of specialized agents adapted from AutoGen’s Magentic-One system. The agents work together to create a modular system: Orchestrator is the lead agent, powered by a large language model (LLM), that performs co-planning with the user, decides when to ask the user for feedback, and delegates sub-tasks to the remaining agents to complete. WebSurfer is an LLM agent equipped with a web browser that it can control. Given a request by the Orchestrator, it can click, type, scroll, and visit pages in multiple rounds to complete the request from the Orchestrator. Coder is an LLM agent equipped with a Docker code-execution container. It can write and execute Python and shell commands and provide a response back to the Orchestrator. FileSurfer is an LLM agent equipped with a Docker code-execution container and file-conversion tools from the MarkItDown (opens in new tab) package. It can locate files in the directory controlled by Magentic-UI, convert files to markdown, and answer questions about them. Figure 6: System architecture diagram of Magentic-UI To interact with Magentic-UI, users can enter a text message and attach images. In response, Magentic-UI creates a natural-language step-by-step plan with which users can interact through a plan-editing interface. Users can add, delete, edit, regenerate steps, and write follow-up messages to iterate on the plan. While the user editing the plan adds an upfront cost to the interaction, it can potentially save a significant amount of time in the agent executing the plan and increase its chance at success. The plan is stored inside the Orchestrator and is used to execute the task. For each step of the plan, the Orchestrator determines which of the agents (WebSurfer, Coder, FileSurfer) or the user should complete the step. Once that decision is made, the Orchestrator sends a request to one of the agents or the user and waits for a response. After the response is received, the Orchestrator decides whether that step is complete. If it is, the Orchestrator moves on to the following step. Once all steps are completed, the Orchestrator generates a final answer that is presented to the user. If, while executing any of the steps, the Orchestrator decides that the plan is inadequate (for example, because a certain website is unreachable), the Orchestrator can replan with user permission and start executing a new plan. All intermediate progress steps are clearly displayed to the user. Furthermore, the user can pause the execution of the plan and send additional requests or feedback. The user can also configure through the interface whether agent actions (e.g., clicking a button) require approval. Evaluating Magentic-UI Magentic-UI innovates through its ability to integrate human feedback in its planning and execution of tasks. We performed a preliminary automated evaluation to showcase this ability on the GAIA benchmark (opens in new tab) for agents with a user-simulation experiment. Evaluation with simulated users Figure 7: Comparison on the GAIA validation set of the accuracy of Magentic-One, Magentic-UI in autonomous mode, Magentic-UI with a simulated user powered by a smarter LLM than the MAGUI agents, Magentic-UI with a simulated user that has a\access to side information about the tasks, and human performance. This shows that human-in-the-loop can improve the accuracy of autonomous agents, bridging the gap to human performance at a fraction of the cost. GAIA is a benchmark for general AI assistants, with multimodal question-answer pairs that are challenging, requiring the agents to navigate the web, process files, and execute code. The traditional evaluation setup with GAIA assumes the system will autonomously complete the task and return an answer, which is compared to the ground-truth answer.  To evaluate the human-in-the-loop capabilities of Magentic-UI, we transform GAIA into an interactive benchmark by introducing the concept of a simulated user. Simulated users provide value in two ways: by having specific expertise that the agent may not possess, and by providing guidance on how the task should be performed. We experiment with two types of simulated users to show the value of human-in-the-loop: (1) a simulated user that is more intelligent than the Magentic-UI agents and (2) a simulated user with the same intelligence as Magentic-UI agents but with additional information about the task. During co-planning, Magentic-UI takes feedback from this simulated user to improve its plan. During co-tasking, Magentic-UI can ask the (simulated) user for help when it gets stuck. Finally, if Magentic-UI does not provide a final answer, then the simulated user provides an answer instead. The simulated user is an LLM without any tools, instructed to interact with Magentic-UI the way we expect a human would act. The first type of simulated user relies on OpenAI’s o4-mini, more performant at many tasks than the one powering the Magentic-UI agents (GPT-4o). For the second type of simulated user, we use GPT-4o for both the simulated user and the rest of the agents, but the user has access to side information about each task. Each task in GAIA has side information, which includes a human-written plan to solve the task. While this plan is not used as input in the traditional benchmark, in our interactive setting we provide this information to the second type of simulated user,which is powered by an LLM so that it can mimic a knowledgeable user. Importantly, we tuned our simulated user so as not to reveal the ground-truth answer directly as the answer is usually found inside the human written plan. Instead, it is prompted to guide Magentic-UI indirectly. We found that this tuning prevented the simulated user from inadvertently revealing the answer in all but 6% of tasks when Magentic-UI provides a final answer.  On the validation subset of GAIA (162 tasks), we show the results of Magentic-One operating in autonomous mode, Magentic-UI operating in autonomous mode (without the simulated user), Magentic-UI with simulated user (1) (smarter model), Magentic-UI with simulated user (2) (side-information), and human performance. We first note that Magentic-UI in autonomous mode is within a margin of error of the performance of Magentic-One. Note that the same LLM (GPT-4o) is used for Magentic-UI and Magentic-One. Magentic-UI with the simulated user that has access to side information improves the accuracy of autonomous Magentic-UI by 71%, from a 30.3% task-completion rate to a 51.9% task-completion rate. Moreover, Magentic-UI only asks for help from the simulated user in 10% of tasks and relies on the simulated user for the final answer in 18% of tasks. And in those tasks where it does ask for help, it asks for help on average 1.1 times. Magentic-UI with the simulated user powered by a smarter model improves to 42.6% where Magentic-UI asks for help in only 4.3% of tasks, asking for help an average of 1.7 times in those tasks. This demonstrates the potential of even lightweight human feedback for improving performance (e.g., task completion) of autonomous agents, especially at a fraction of the cost compared to people completing tasks entirely manually.  Learning and reusing plans As described above, once Magentic-UI completes a task, users have the option for Magentic-UI to learn a plan based on the execution of the task. These plans are saved in a plan gallery, which users and Magentic-UI can access in the future. The user can select a plan from the plan gallery, which is displayed by clicking on the Saved Plans button. Alternatively, as a user enters a task that closely matches a previous task, the saved plan will be displayed even before the user is done typing. If no identical task is found, Magentic-UI can use AutoGen’s Task-Centric Memory (opens in new tab) to retrieve plans for any similar tasks. Our preliminary evaluations show that this retrieval is highly accurate, and when recalling a saved plan can be around 3x faster than generating a new plan. Once a plan is recalled or generated, the user can always accept it, modify it, or ask Magentic-UI to modify it for the specific task at hand.  Safety and control Magentic-UI can surf the live internet and execute code. With such capabilities, we need to ensure that Magentic-UI acts in a safe and secure manner. The following features, design decisions, and evaluations were made to ensure this: Allow-list: Users can set a list of websites that Magentic-UI is allowed to access. If Magentic-UI needs to access a website outside of the allow-list, users must explicitly approve it through the interface Anytime interruptions: At any point of Magentic-UI completing the task, the user can interrupt Magentic-UI and stop any pending code execution or web browsing. Docker sandboxing: Magentic-UI controls a browser that is launched inside a Docker container with no credentials, which avoids risks with logged-in accounts and credentials. Moreover, any code execution is also performed inside a separate Docker container to avoid affecting the host environment in which Magentic-UI is running. This is illustrated in the system architecture of Magentic-UI (Figure 3). Detection and approval of irreversible agent actions: Users can configure an action-approval policy (action guards) to determine which actions Magentic-UI can perform without user approval. In the extreme, users can specify that any action (e.g., any button click) needs explicit user approval. Users must press an “Accept” or “Deny” button for each action. In addition to the above design decisions, we performed a red-team evaluation of Magentic-UI on a set of internal scenarios, which we developed to challenge the security and safety of Magentic-UI. Such scenarios include cross-site prompt injection attacks, where web pages contain malicious instructions distinct from the user’s original intent (e.g., to execute risky code, access sensitive files, or perform actions on other websites). It also contains scenarios comparable to phishing, which try to trick Magentic-UI into entering sensitive information, or granting permissions on impostor sites (e.g., a synthetic website that asks Magentic-UI to log in and enter Google credentials to read an article). In our preliminary evaluations, we found that Magentic-UI either refuses to complete the requests, stops to ask the user, or, as a final safety measure, is eventually unable to complete the request due to Docker sandboxing. We have found that this layered approach is effective for thwarting these attacks. We have also released transparency notes, which can be found at: https://github.com/microsoft/magentic-ui/blob/main/TRANSPARENCY_NOTE.md (opens in new tab) Open research questions  Magentic-UI provides a tool for researchers to study critical questions in agentic systems and particularly on human-agent interaction. In a previous report (opens in new tab), we outlined 12 questions for human-agent communication, and Magentic-UI provides a vehicle to study these questions in a realistic setting. A key question among these is how we enable humans to efficiently intervene and provide feedback to the agent while executing a task. Humans should not have to constantly watch the agent. Ideally, the agent should know when to reach out for help and provide the necessary context for the human to assist it. A second question is about safety. As agents interact with the live web, they may become prone to attacks from malicious actors. We need to study what necessary safeguards are needed to protect the human from side effects without adding a heavy burden on the human to verify every agent action. There are also many other questions surrounding security, personalization, and learning that Magentic-UI can help with studying.  Conclusion Magentic-UI is an open-source agent prototype that works with people to complete complex tasks that require multi-step planning and browser use. As agentic systems expand in the scope of tasks they can complete, Magentic-UI’s design enables better transparency into agent actions and enables human control to ensure safety and reliability. Moreover, by facilitating human intervention, we can improve performance while still reducing human cost in completing tasks on aggregate. Today we have released the first version of Magentic-UI. Looking ahead, we plan to continue developing it in the open with the goal of improving its capabilities and answering research questions on human-agent collaboration. We invite the research community to extend and reuse Magentic-UI for their scientific explorations and domains.  Opens in a new tab
    44 Comentários 0 Compartilhamentos
  • Coinbase cyberattack: What users need to know about stolen customer data, password security, and more in $400 million incident

    Stock in Coinbase, the largest U.S.-based cryptocurrency exchange, fell more than 8% Thursday on news it was the victim of a cyberattack, in which hackers successfully bribed overseas contractors to leak important information so they could steal customer data. The company estimates it could cost million to resolve the situation.

    While investors may be concerned, Coinbase customers undoubtedly are as well. Here’s what users of the crypto exchange need to know.

    What happened?

    Coinbase reported in a Securities and Exchange Commissionfiling that on May 11, it received an email from an entity claiming to have obtained information about certain Coinbase customer accounts and internal Coinbase documentation—including materials relating to customer-service and account-management systems.

    The filing said hackers sent an email threatening to publish customers’ personal data if Coinbase did not pay a million ransom, which CEO Brian Armstrong confirmed on X was specifically for “million in Bitcoin.” According to the SEC filing, Coinbase learned the cybercriminals obtained the data by paying off multiple overseas contractors or employees working in support roles. Once detected, Coinbase immediately terminated those involved.

    Coinbase said it did not pay the ransom and has been working with law enforcement to investigate the breach. It’s establishing a million reward for information leading to the arrest and conviction of those responsible for the attack.

    Was my Coinbase password or private key leaked in the attack?

    No. The SEC filing said the data breach did not compromise customer passwords or private keys.

    Were my Coinbase funds exposed in the attack?

    According to the SEC filing, neither “targeted contractors” nor “employees” were able to access customer funds.

    What about Coinbase customer data like my email, address, and phone number?

    Yes, according to Coinbase’s blog, the following personal information was compromised:

    Name, address, phone, and email

    Masked Social SecurityMasked bank‑account numbers and some bank account identifiers

    Government‑ID imagesAccount dataLimited corporate dataHow can I protect myself?

    Coinbase told Fast Company: “Expect impostors. Scammers—related to this incident or not—may pose as Coinbase employees and try to pressure you into moving your funds.”

    Additionally, the company outlined what customers can do in this post.

    What should I do if I receive a phone call, text, or request from Coinbase?

    A Coinbase spokesperson told Fast Company: “If you receive this call, hang up the phone. Coinbase will never ask you to contact an unknown number to reach us.”

    Again, remember, Coinbase will never call or text, or ask for your password or two-factor authenticationcodes, or for you to transfer assets to a specific or new address, account, vault, or wallet.

    I think my Coinbase information was leaked in the cyberattack. What should I do?

    Coinbase said it will reimburse customers who were tricked into sending funds to the attacker due to social engineering attacks.

    If your data was accessed, you should have already received an email; notifications were sent Wednesday, May 15, at 7:20 a.m. ET to affected customers.

     Flagged accounts now require additional ID checks on large withdrawals and include mandatory scam‑awareness prompts. As Coinbase monitors high-risk transactions, customers may experience delays.

    The company said it is opening a new support hub in the U.S., adding stronger security controls and monitoring across all locations, and will keep the community updated as the investigation progresses.
    #coinbase #cyberattack #what #users #need
    Coinbase cyberattack: What users need to know about stolen customer data, password security, and more in $400 million incident
    Stock in Coinbase, the largest U.S.-based cryptocurrency exchange, fell more than 8% Thursday on news it was the victim of a cyberattack, in which hackers successfully bribed overseas contractors to leak important information so they could steal customer data. The company estimates it could cost million to resolve the situation. While investors may be concerned, Coinbase customers undoubtedly are as well. Here’s what users of the crypto exchange need to know. What happened? Coinbase reported in a Securities and Exchange Commissionfiling that on May 11, it received an email from an entity claiming to have obtained information about certain Coinbase customer accounts and internal Coinbase documentation—including materials relating to customer-service and account-management systems. The filing said hackers sent an email threatening to publish customers’ personal data if Coinbase did not pay a million ransom, which CEO Brian Armstrong confirmed on X was specifically for “million in Bitcoin.” According to the SEC filing, Coinbase learned the cybercriminals obtained the data by paying off multiple overseas contractors or employees working in support roles. Once detected, Coinbase immediately terminated those involved. Coinbase said it did not pay the ransom and has been working with law enforcement to investigate the breach. It’s establishing a million reward for information leading to the arrest and conviction of those responsible for the attack. Was my Coinbase password or private key leaked in the attack? No. The SEC filing said the data breach did not compromise customer passwords or private keys. Were my Coinbase funds exposed in the attack? According to the SEC filing, neither “targeted contractors” nor “employees” were able to access customer funds. What about Coinbase customer data like my email, address, and phone number? Yes, according to Coinbase’s blog, the following personal information was compromised: Name, address, phone, and email Masked Social SecurityMasked bank‑account numbers and some bank account identifiers Government‑ID imagesAccount dataLimited corporate dataHow can I protect myself? Coinbase told Fast Company: “Expect impostors. Scammers—related to this incident or not—may pose as Coinbase employees and try to pressure you into moving your funds.” Additionally, the company outlined what customers can do in this post. What should I do if I receive a phone call, text, or request from Coinbase? A Coinbase spokesperson told Fast Company: “If you receive this call, hang up the phone. Coinbase will never ask you to contact an unknown number to reach us.” Again, remember, Coinbase will never call or text, or ask for your password or two-factor authenticationcodes, or for you to transfer assets to a specific or new address, account, vault, or wallet. I think my Coinbase information was leaked in the cyberattack. What should I do? Coinbase said it will reimburse customers who were tricked into sending funds to the attacker due to social engineering attacks. If your data was accessed, you should have already received an email; notifications were sent Wednesday, May 15, at 7:20 a.m. ET to affected customers.  Flagged accounts now require additional ID checks on large withdrawals and include mandatory scam‑awareness prompts. As Coinbase monitors high-risk transactions, customers may experience delays. The company said it is opening a new support hub in the U.S., adding stronger security controls and monitoring across all locations, and will keep the community updated as the investigation progresses. #coinbase #cyberattack #what #users #need
    WWW.FASTCOMPANY.COM
    Coinbase cyberattack: What users need to know about stolen customer data, password security, and more in $400 million incident
    Stock in Coinbase (COIN), the largest U.S.-based cryptocurrency exchange, fell more than 8% Thursday on news it was the victim of a cyberattack, in which hackers successfully bribed overseas contractors to leak important information so they could steal customer data. The company estimates it could cost $400 million to resolve the situation. While investors may be concerned, Coinbase customers undoubtedly are as well. Here’s what users of the crypto exchange need to know. What happened? Coinbase reported in a Securities and Exchange Commission (SEC) filing that on May 11, it received an email from an entity claiming to have obtained information about certain Coinbase customer accounts and internal Coinbase documentation—including materials relating to customer-service and account-management systems. The filing said hackers sent an email threatening to publish customers’ personal data if Coinbase did not pay a $20 million ransom, which CEO Brian Armstrong confirmed on X was specifically for “$20 million in Bitcoin.” According to the SEC filing, Coinbase learned the cybercriminals obtained the data by paying off multiple overseas contractors or employees working in support roles. Once detected, Coinbase immediately terminated those involved. Coinbase said it did not pay the ransom and has been working with law enforcement to investigate the breach. It’s establishing a $20 million reward for information leading to the arrest and conviction of those responsible for the attack. Was my Coinbase password or private key leaked in the attack? No. The SEC filing said the data breach did not compromise customer passwords or private keys. Were my Coinbase funds exposed in the attack? According to the SEC filing, neither “targeted contractors” nor “employees” were able to access customer funds. What about Coinbase customer data like my email, address, and phone number? Yes, according to Coinbase’s blog, the following personal information was compromised: Name, address, phone, and email Masked Social Security (last 4 digits only) Masked bank‑account numbers and some bank account identifiers Government‑ID images (e.g., driver’s license, passport) Account data (balance snapshots and transaction history) Limited corporate data (including documents, training materials, and communications available to support agents) How can I protect myself? Coinbase told Fast Company: “Expect impostors. Scammers—related to this incident or not—may pose as Coinbase employees and try to pressure you into moving your funds.” Additionally, the company outlined what customers can do in this post. What should I do if I receive a phone call, text, or request from Coinbase? A Coinbase spokesperson told Fast Company: “If you receive this call, hang up the phone. Coinbase will never ask you to contact an unknown number to reach us.” Again, remember, Coinbase will never call or text, or ask for your password or two-factor authentication (2FA) codes, or for you to transfer assets to a specific or new address, account, vault, or wallet. I think my Coinbase information was leaked in the cyberattack. What should I do? Coinbase said it will reimburse customers who were tricked into sending funds to the attacker due to social engineering attacks. If your data was accessed, you should have already received an email; notifications were sent Wednesday, May 15, at 7:20 a.m. ET to affected customers.  Flagged accounts now require additional ID checks on large withdrawals and include mandatory scam‑awareness prompts. As Coinbase monitors high-risk transactions, customers may experience delays. The company said it is opening a new support hub in the U.S., adding stronger security controls and monitoring across all locations, and will keep the community updated as the investigation progresses.
    0 Comentários 0 Compartilhamentos