A Berlin court has raised questions over whether data from 120 million messages obtained by police hacking an encrypted phone service can continue to be lawfully used as evidence in prosecutions in Germany and other European countries.The Landgericht Berlin Regional Court has ruled that text messages intercepted by French police from the EncroChat encrypted phone network cannot be used to prosecute a suspect for alleged drugs trafficking offences in Germany.The decision, by Germanys largest criminal court, calls into question previous assumptions that under Europes mutual recognition principle, intercept evidence obtained by one member state can automatically be used as evidence in other European states.The court ruling is likely to have implications for the use of evidence obtained from future law enforcement hacking operations into encrypted communications systems, defence lawyer Christian Ldden told Computer Weekly.EncroChat is one of a series of encrypted phone and messaging services to be infiltrated by collaborating law enforcement agencies across Europe since 2020, sparking prosecutions of organised crime groups for drug trafficking and money laundering in multiple countries.French and Dutch police harvested messages from 4,600 EncroChat phone users in Germany and tens of thousands of phone users in other countries after infiltrating EncroChat servers hosted at the OVH datacentre in Roubaix, France, in a novel hacking operation in 2020.A three-year investigation by police into organised crime and drug groups using EncroChat phones led to 6,500 arrests worldwide and the seizure of nearly 900min cash and assets.The lawfulness of the use of hacked data from EncroChat and other encrypted phone networks has now been called into question following a ruling by the Berlin Regional Court.The months-long trial heard evidence from German investigators and prosecutors, and reviewed translations of evidence disclosed by the UKs National Crime Agency during criminal trials involving EncroChat in the UK.A grand chamber of the Berlin Regional Court, made up of three professional judges, a presiding judge and two lay people, found in an oral decision in December that contrary to arguments by European prosecutors, French investigators had not intercepted EncroChat data from a central server in France, but had harvested it from the handsets of EncroChat users in German territory.Read more about key EncroChat decisions in GermanyOctober 2022: Berlins Regional Court has asked the European Court of Justice to answer questions about whether the use of hacked EncroChat phone evidence complies with European law.July 2023: The European Court of Justice hears evidence on whether the collection and sharing of data intercepted by law enforcement from the EncroChat crypto phone network is compatible with European law.October 2023: Germany lawfully obtained data on German EncroChat users from France, but whether the evidence is legally admissible is a matter for national courts, says advocate general.April 2024: The Court of Justice of the European Union (CJEU) rules that EU member states must formally notify other member states when they intercept communications in their jurisdiction, opening up the way for further legal challenges.Under German law, that meant prosecutors were obliged to seek approval from the German courts to use the French-supplied data in Germany.However, the presiding judge found that prosecutors had failed to seek judicial approval and that German courts would not have authorised the hacking operation against EncroChat under German law.The decision came after the Berlin Regional Court submitted questions to the Court of Justice of the European Union (CJEU) asking whether Frances sharing of hacked EncroChat messages with Germany was permitted under European law.The European court found that, under the European Invesitigation Order (EIO) Directive, France should have formally notified Germany of the interception of EncroChat phones on German soil, and given German authorities the opportunity to object to the operation within 96 hours, if they wished.The court of justice found, contrary to previous German court decisions, that the protections offered by Article 31 of the EIO Directive were designed to protect the rights not only of the country receiving evidence from another EU state but also the individual users of telecoms services intercepted by law enforcement.That contradicted earlier findings of the German supreme court that found Article 31 exists only to support the sovereignty of member states, and cannot be claimed by German citizens as a measure to protect their rights.Following the CJEUs decision, the Berlin Regional Court found in its latest ruling that the principle of mutual confidence in actions of other member states during judicial cooperation only meant Germany should recognise that Frances actions were legal under French law.The presiding judge, Kristin Klimke, found that the German court still had a duty to examine whether the French operation against EncroChat would be legal under German law. And in this case, a German court would not have approved the operation under German law because the evidence of suspicion did not meet the threshold to justify an equivalent hacking operation in Germany.The judge also found that prosecutors had not established that evidence of serious crimes could not have been obtained in a less obtrusive way than by intercepting the data of all EncroChat phone users in Germany.The principle of European cooperation is not intended to require each national law authority to adopt the same criteria for conducting state hacking operations, but is intended to enable cooperation between countries with different laws to protect the privacy and other rights of their citizens, the judge found.Although the Court of Justice of the European Union allowed German prosecutors to request EncroChat data from France, the CJEU did not go on to say that prosecutors could use the data without approval from a German court.In another legally significant decision, the judge found that the hacking operation against EncroChat was not simply a French police operation but was a joint European operation involving a number of other EU member states.France went beyond surveilling the 300 French users of EncroChat, gathering data from all EncroChat users in Europe, the judge found. France had notified its partner countries in advance of the hacking operation.However French prosecutors failed to comply with European law by failing to follow the correct procedures under EU law to inform Germany of its plans to obtain the phone data of German citizens.Frances notification should have contained details of the targets identified by phone number, IP address or email, the identity of individuals targeted, including their address, date of birth and social security numbers, as well as a description of the offence committed.The Berlin Regional Court also found that the French authorities had not disclosed their communications with German police and that no information had been supplied to the court on how the data had been intercepted raising questions over whether defendants had adequate information to challenge the validity of the data.German defence lawyer Christian Ldden, who is a member of an international group of lawyers collaborating on EncroChat and similar cases, said the court was the first to try to understand what happened before and during the EncroChat operation. The judge found that Germany, rather than simply taking data France had already obtained from EncroChat, had been informed about the hacking operation in advance and had therefore participated in the operation.At the end of the day, she said that under German and European law, the evidence is not allowed to be used in court, he added.Ldden said the decision would set a precedent for other cases heard in Germany, though courts elsewhere would make their own decisions on the admissibility of EncroChat evidence. The case is also likely to impact the use of intercept evidence in other cases in Europe, he said.Dutch defence lawyer Justus Reisinger said the Berlin courts decision could have massive implications for cases in Holland.This decision basically confirms our defence arguments in the Netherlands from the recent year. Previously, the Supreme Court rejected my arguments on this point, but along with the Berlin court, even academics are saying that an interpretation like that from the Dutch Supreme Court cant stand. So a legal landslide is quite possible and justified, he said.Bojana Franovi, a lawyer in Montengro dealing with evidence from police hacking of Sky ECC and the FBI-run Anom encrypted phone network, said the decision was likely to influence judicial decisions in her country.Italian lawyer, Daniel Fiorino, said that the Berlin court decision as an excellent result but described the legal situation in Italy as very complex.We have numerous trials still underway, he said.Everyone in the judiciary, at least in Montenegro, is very keen on what the other countries are doing and how they are dealing with those cases, she said.A final written version of the decision has yet to be published.Prosecutors are expected to appeal the decision to the Supreme Court in Germany.The Berlin Regional Court ruled that EncroChat data cannot be used in evidence in a criminal trial.Although data from EncroChat phones was obtained lawfully under French law, a German court is still required to decide whether the interception measures taken by France were permissible under German law.Under German law, the suspicion that users of EncroChat were committing crimes did not reach the threshold to justify intercepting all EncroChat communications.The principle of mutual cooperation between European member states must recognise national measures to protect citizens fundamental rights in cooperating countries.Although the European Court of Justice concluded that German prosecutors were permitted to request EncroChat data from France, that does not in itself mean prosecutors could also use the data in prosecutions.It was not established that evidence against suspects could not have been gathered by less draconian means other than by than intercepting their communications.Germanys legal role in EncroChat2017: French investigators establish that EncroChat encrypted phones were used in a number of drugs related offences.2018: French investigators copy data from an EncroChat server at the OVH datacentre in Roubaix, France. The server data reveals that over 66,000 SIM cards are registered on EncroChat. Investigators are able to decrypt 3,500 files including encrypted notes made by phone users.2018: The German Federal Criminal Police Office (BKA) discovers that EncroChat phones are being used in Germany in serious crimes.30 January 2020: A court in Lille, France, approves the use of a data interception device on the EncroChat server and on EncroChat handsets.2020: The German BKA and the Central Office for Combating Cyber Crime (ZIT), located at the Public Prosecutors Office in Frankfurt (GSta), begin discussions on an investigation into EncroChat.9 March 2020: The BKA and the ZIT take part in a video conference organised by Eurojust in the Hague with representatives of other countries to discuss how to exploit EncroChat data with the French and Dutch Joint Investigation Team (JIT) working on the hacking operation.13 March 2020: The BKA begins a preliminary investigation into unknown users of the EncroChat service on suspicion that EncroChat phone users were in small-scale trafficking of narcotics and organised crime.20 March 2020: The Lille court in France approves an order to redirect data streams on the EncroChat server to enable the capture of EncroChat data.27 March 2020: The BKA receives a message in English from the French and Dutch JIT through Europols Siena communications systems inviting police authorities in other countries to receive messages from EncroChat. The note asked participating countries to confirm they had been informed of the methods used to obtain data from devices in their jurisdiction. Participating countries could only use the intercept material in investigations after being granted permission by the JIT.27 March 2020: The BKA issues the approvals and confirmations requested by Europol after consulting with GSta Frankfurt.1 April 2020: The French and Dutch JIT installs Trojan Horse or implant software on an EncroChat server hosted in the OVH datacentre in Roubaix, France, which goes live. 3 April 2020: The BKA begins downloading EncroChat data supplied by France, through Europol.7 April 2020: The French investigation is expanded from an investigation into the illegal supply of encryption technology in France to include illegal trade in drugs and weapons offences.1 May 2020: The Lille court in France extends permission to continue technical measures against EncroChats infrastructure for one month.13 May 2020: The BKA writes to the French public prosecutor asking permission to make judicial decisions on the collection of location data and other investigative measures on suspects, without disclosing the French investigation into EncroChat.1 June 2020: The Lille court extends permission to continue technical measures against EncroChats infrastructure for a further four months.2 June 2020: The German public prosecutors office of Frankfurt issues a European Investigation Order formally requesting permission to use the EncroChat data obtained by France in prosecutions.13 June 2020: The Lille court approves Germanys European Investigation Order, giving consent to the use of the data by Germany for judicial investigations and prosecutions.28 June 2020: EncroChat administrators succeed in closing down the EncroChat network after having discovered the hacking operation.9 September 2020: The Public Prosecutors Office in Frankfurt issues a further European Investigation Order for additional data from the French EncroChat operation.2 July 2021: The Public Prosecutors Office in Frankfurt issues a third further European Investigation Order for additional data from the French EncroChat operation.1 July 2021: The Berlin Regional Court finds that EncroChat messages cannot be used in German criminal proceedings.5 July 2021: The Berlin Public Prosecutors Office issues a complaint seeking to overturn the Berlin Regional Courts decision and requesting the reopening of criminal proceedings against the defendant.31 August 2021: The Superior Court in Berlin rules that messages intercepted by French police from the EncroChat encrypted phone network can be used as evidence.2 March 2022: Germanys Supreme Court, the Federal Court of Justice (BGH), rules that EncroChat evidence provided by France to Germany could be used as evidence in Germany for investigating serious criminal offences.October 2022: Berlins Regional Court asks the European Court of Justice to answer questions about whether the use of hacked EncroChat phone evidence complies with European law.July 2023: The European Court of Justice hears evidence on whether the collection and sharing of data intercepted by law enforcement from the EncroChat cryptophone network is compatible with European law.October 2023: Germany lawfully obtained data on German EncroChat users from France, but whether the evidence is legally admissible is a matter for national courts, says the advocate general.April 2024: The CJEU rules that EU member states must formally notify other member states when they intercept communications in their jurisdiction, opening up the way for further legal challenges.December 2024: The Berlin Regional Court finds that EncroChat data supplied by France cannot be lawfully used as evidence in a long-running drugs trafficking case.

ZDNETThe buzzword of the past two years has been AI, and as a result, many of the products at CES featured the technology, or at least claimed to, making it more challenging than ever to choose the best in the category. When executed correctly, integrating AI into consumer tech can significantly improve how helpful users' workflows and lives are by unlocking a new range of possibilities, and that's what ZDNET was on the lookout for at CES.Also: CES 2025: The most impressive products we've seen so farInformed by hands-on time on the show floor, as well as demos of the most cutting-edge AI features in existing hardware, ZDNET rounded up the AI features and products that either are the most likely to transform your life today or have the promise to bring meaningful change in the near future. Keep reading below to find the picks, which will be updated every day with the latest selections. 1. Halliday AI Glasses with Invisible Display Jada Jones/ZDNET The Halliday smart glasses look like any other ordinary glasses frames, light, and compact. However, when worn, in the upper right-hand corner, you can see a display that shows you information like the date, time, translations, messages, and text. The coolest part is its "invisible display" which simply means that what you see isn't coming form the actual displays or lenses, rather an advanced optical module.Also: I tried Halliday's AI smart glasses and they beat my Ray-Ban Metas in key waysFor example, the unit I used to demo the feature had no lenses on it, just the glasses frames, and I could still see what was being shown to me in the demo with ease. At the event in which they were shown, I was also wearing my Ray-Ban Metas, and in comparison, these were noticeably lighter, more akin to my everyday glasses. They can be reserved online for $369 as part of a Kickstarter launch. Show more 2. Natura Humana HumanPods Sabrina Ortiz/ZDNET In the intro to this article, I mentioned some products made the roundup because of the promise they offered, and the Natura Humana HumanPods fall under that category. The idea is simple: you would wear one or two earbuds, up to your preference, all day, double-tapping to access an arsenal of "AI People" LLM-based AI agents with unique personalities and voices. Users can also access the leading LLMs if they choose to, such as ChatGPT, Claude, and more.Before the demo, I was skeptical because I couldn't spot the difference between this product and just using earbuds withChatGPT's Advanced Voice. However, the experience is truly much more seamless than that of existing options, as all you have to do is tap on your ear to talk to your AI person. Even in the loud expo room, my conversation with "Nature," one of the AI people, was seamless. The open form factor of the earbuds is also very comfortable, leveraging a unique Gravity Hook to stay in place comfortably. It is not available for consumers yet, but there is awaitliston its site. Show more 3. Roborock Saros Z70 Maria Diaz/ZDNET A good rule of thumb at CES is that if there is a lot of buzz among the tech aficionados in attendance about a product, it is probably for good reason. The Roborock Saros 270 falls under that category, with its mechanical arm that removes obstacles in its path causing most people to stop in their tracks.Using the same AI-powered camera that helps robots avoid objects, the robovac can identify objects in the way and use it claw-like arm can remove obstacles under 300 grams, or about 8 oz, using OmniGrip technology. Some products it can move include socks, napkins, tissues, and even sandals. Once it picks up the item, it can then clean underneath it like it regularly would if its path were unobstructed. Beyond the mechanical arm, it is a very capable robot vacuum with 22,000Pa of suction. Show more 4. Gemini integration into Google TV Sabrina Ortiz/ZDNET I can sense the hesitation from readers before even publishing it, but after seeing the demo in action, I am positive it is an upgrade all Google TV users will be thankful for. With the Gemini integration, you'll now be able to speak to it more seamlessly, using natural language and conversing with Google TV. This also means your queries can be more complex, ranging far beyond anything Google Assistant could do before.For example, in my demo, the user asked more involved questions such as, "Tell me about the solar system for a third grader," and "Help me plan a vacation spot with great beaches." In both instances, Gemini was able to provide an answer and present it in a visual way that made sense for a TV, including links to YouTube videos. Show more 4. Mudra Link Sabrina Ortiz/ZDNET If you have ever seen the hand gestures on the Apple Vision Pro and wished you can control everything that way, with the Mudra Link you can. The Neural Wristband leverages proprietary sensors and AI algorithms to allow users to interact with their devices hands-free, using wrist movements and hand gestures. Although it may seem like it fits more into the realm of XR/VR, seven years of AI developments power the tech, according to the demoer.The wristband fits comfortable on your wrist and pairs to your device the same way a mouse would, utilizing Bluetooth. During my demo, I was able to nearly instantly get the hang of using it, slicing fruit in Fruit Ninja like a natural. The Mudra Band was even awarded a CES 2025 Innovation Award in the XR Technologies & Accessories category. It retails for $299, and is available for purchase right now for $199. Show more 5. Vasco Translator Q1 Sabrina Ortiz/ZDNET Vasco isn't new to translating devices, with the Vasco Translator E1 demoed last year at CES leaving ZDNET impressed. Since last year, AI has facilitated the release of many more AI translators, making it harder to stand out. However, Vasco stepped up to the plate.It's new Vasco Translator Q1 builds on its predecessor, featuring a new real-time Call Translator, which can translate calls in real-time in up to 50 languages, and the star of the show, its voice cloning features which allow users to make a digital copies of their voices for more seamless conversations.I demoed the feature and was impressed at how similar the voice sounded to mine from only reading one sentence. This seems like a more novel and useful application of AI, helping bridge the disconnect that happens in conversations when using a robotoic sounding AI translator. The Vasco Translator Q1 is not yet available for purchase, with market release scheduled for Q2 2025. Show more 5. Amp Sabrina Ortiz/ZDNET Applications of AI in everyday life don't have to be limited to language, and the new Amp strength machine is a great example. The souped up cable machine not only has a beautiful sleek look that differentiates it from traditional workout equipment, but features AI to upgrade your workout experience.Using the camera and AI, it offers personalized coaching with smart tracking. On the app, you can select what parts of your body you'd like to workout, and using your data and AI the device can create a custom routine for you.It has a business model similar to Peloton, with a hardware cost of $1,795 which gets you the device, two ankle straps, a dual handle, a single handle, rope, and T-bar, and a monthly subscription cost there after of $23 a month. Through the subscription you get to access all of the AI enhancements. It is available for reserve now, scheduled to ship in early 2025. Show more View now at Ampfit 6. Bee AI-wearable Sabrina Ortiz/ZDNET It is difficult to find an implementation of AI in a wearable that is truly unique, but I have never heard of anything like the Bee AI-wearable. The wristband listens to your conversations all day, unless manually paused with the button on it, and uses that information to get to know you, provide AI summaries of your conversations, transcripts, and actionable insights.Of course, it wouldn't be an AI product if it didn't feature a chatbot, which you can use to chat with to learn more about anything that happened in your day, such as referring to what someone said in an earlier conversation, or even how to improve your own behaviors. It can also be integrated with third parties like Google Calendar and Gmail.The battery lasts seven days, and it feels as comfortable as any other wristband device can. Of course, being listened to 24/7 isn't for everyone, but Bee reassures users that users are in control of their memories, no human can see the data, the data will not be sold or trained on, and the audio is not saved. The wearable costs $50 and is available for purchase now for iOS only, with an Android offering coming later this month before the official launch. Show more 7. Omnia Smart Mirror Sabrina Ortiz/ZDNET Even though this mirror may just seem like an aesthetic decorative mirror, it holds much more. The conceptual mirror, developed by Withings, is a smart scale with 360-degree health screenings, including a person's heart health, nutrition, body composition, lung function, activity, and sleep. It uses AI to develop meaningful insights based on your inputted data, as well as the data collected.Although this is a concept and not likely to unveil anytime soon, it is an exciting look at what AI will be able to do for the health and wellness space. Some features, such as the AI-powered assistant, will be coming later this year within the Withings+ services. Show more FAQ What is CES? The Consumer Electronics Show (CES) is an industry trade show that features the biggest tech players from across the globe, showcasing their latest consumer technology with daily product launches, keynotes, activations, and demos. Show more When is CES 2025? The official days of CES 2025 are Tuesday, Jan. 7, to Friday, Jan. 10. However, since Jan. 5 and Jan. 6 are media days, a lot of the news breaks before the conference even starts. Show more CES 2025

With a display that's built into the frame - not the lens - these Halliday smart glasses are making quite a splash at CES 2025.

This is the perfect tool for blanketing a large home with high-speed Wi-Fi, but it will come at a cost.

Mother hippo Iris and baby Poppy take a dip in a shallow pool.Metro Richmond ZooIf you happened to be a pygmy hippo born in 2024, then youre an internet superstar. The world first went bonkers for Moo Deng, a tiny hippo born in a Thailand zoo. Next came Haggis, a wee hippo born in Scotland. The latest is a newbie at the Metro Richmond Zoo in Virginia. The zoo announced the results of a naming poll this week. Please give a warm welcome to Poppy, the baby pygmy hippopotamus.Pygmy hippo gets a nameIt took two rounds of voting to arrive at the babys final name. The four contenders were Poppy (for the flower), Juniper (for the evergreen), Hammie Mae (for Virginia ham) and Omi (water in the West African language Yoruba). Poppy and Hammie Mae were the top vote-getters in the first round. Poppy reigned supreme in the runoff vote. Poppy is the perfect name for this playful, perky, and ever-increasing popular hippo, the zoo said in a statement. Her name is a flower just like her mother Iris.Over 116,000 votes came in from 165 countries. Poppy squeaked past the food-themed Hammie Mae with 52.8% of the vote. Momma hippo Iris even got a vote. The zoo set up two piles of food with the names Poppy and Hammie Mae above them. Iris lumbered over and started eating from the pile under Poppy. The Metro Richmond Zoo shared a video to social media on Jan. 7 showing Iris voting.MORE FOR YOUPygmy hippo celebritiesPoppy is the latest in a lineup of baby pygmy hippos that have attracted a lot of attention. Thailands Moo Deng became popular in 2024 after becoming the subject of memes, TikTok makeup tutorials and even a Saturday Night Live controversy. Moo Dengs ascendance turned a spotlight on pygmy hippos in general. Adult pygmy hippos can reach 600 pounds. Thats still much smaller than a standard hippo.Poppy was born on Dec. 9. She represents hope for a species under duress in the wild from hunting and loss of habitat. Pygmy hippos (Choeropsis liberiensis) are native to West Africa. Pygmy hippos are classified as endangered on the International Union for Conservation of Natures Red List of Threatened Species. Fewer than 2,500 mature individuals still live in the wild.The Metro Richmond Zoo shared a video of Poppy taking a splash in a shallow pool with her mom.Poppy flips her ears, wiggles around and inspects the camera. Its easy to see why the babies of this species get so much attention from humans. They come across as curious, sassy and cute. Their small size makes them much less intimidating than regular hippos, which have a reputation as dangerous, fast-moving animals.Poppy is a celebrity that isnt aware of her celebrity status. Shes busy doing baby-hippo things. Chowing down is a top priority. She will continue learning how to be a pygmy hippo from mom, the zoo said. She has already started nibbling on some solid foods, and will only eat more and more each day. Visitors to the zoo can see water-loving Poppy and Iris in their indoor pool area. Eventually, the baby will get to explore her larger outdoor enclosure, no doubt to the delight of her adoring fans.

Self-driving cars, specialized robots for medical procedures or factory work, and general-purpose robots as home assistants will be the next major shift for industries.

ReacherCredit: AmazonWeve seen a teaser for Reachers third season, headed to Prime Video this coming February. But now we have the first official trailer and it has renewed my hope in this show, after a truly abysmal second season.Season 2 was a huge disappointmentI included it in my Worst TV Shows Of 2024 listafter the debut season was a close to perfect adaptation of the first Lee Child Reacher novel, Killing Floor. Beyond its many plot holes and lackluster plot, the second season saw Reacher team up with his ex-army buddies, which kind of defeats the entire point of Reacher being a one-man army who can take down even the most intimidating foes all by his lonesome. Details matter, as the characters in Season 2 never stopped reminding us.Dont get me wrong: Reacher almost always has help in some form or another. He had help from Roscoe and Finlay in Season 1, but they werent fighting alongside him. The fight scenes in Season 2 were almost all terrible by comparison.Season 3 looks like a return to form. Heres the trailer:This season is based on the seventh Reacher novel, Persuader. Of the books in the series that Ive read, this is my second favorite after Killing Floor, and one of the better ones I can think of to adapt for TV. I wont spoil what happens in the book, but the official logline for the season reads: In the third season of the action-packed series, Reacher (Alan Ritchson) hurtles into the dark heart of a vast criminal enterprise when trying to rescue an undercover DEA informant whose time is running out. There he finds a world of secrecy and violenceand confronts some unfinished business from his own past.Reacher will team up with DEA Susan Duffy (Sonya Cassidy) who, if the story follows the book, will also be our leading mans romantic interest this time around. Reacher will go up against a cabal of criminals including Paulie (Olivier Richters) pictured in the top image of this post. As you can see, hes quite a lot bigger than Reacher. Hes also a very bad dude. Anthony Michael Hall also joins the cast as Zackary Beck, the chief antagonist of the season.Read More: Apple TV Has Stolen HBOs Thunder In The Streaming EraThe first three episodes of Reacher drop on Thursday, February 20th with subsequent episodes landing on Prime Video each week. I will be reviewing here on this blog. Fingers crossed that this season lives up to the first and that writer and showrunner Nick Sanotra learned from the second seasons mistakes.

By Harry BoothJanuary 8, 2025 12:46 PM ESTIn 2017, Beijing unveiled an ambitious roadmap to dominate artificial intelligence development, aiming to secure global leadership by 2030. By 2020, the plan called for iconic advances in AI to demonstrate its progress. Then in late 2022, OpenAIs release of ChatGPT took the world by surpriseand caught China flat-footed.At the time, leading Chinese technology companies were still reeling from an 18-month government crackdown that shaved around $1 trillion off China's tech sector. It was almost a year before a handful of Chinese AI chatbots received government approval for public release. Some questioned whether Chinas stance on censorship might hobble the countrys AI ambitions. Meanwhile, the Biden administrations export controls, unveiled just a month before ChatGPTs debut, aimed to cut China off from the advanced semiconductors essential for training large-scale AI models. Without cutting-edge chips, Beijings goal of AI supremacy by 2030 appeared increasingly out of reach.But fast forward to today, and a flurry of impressive Chinese releases suggests the U.S.s AI lead has shrunk. In November, Alibaba and Chinese AI developer DeepSeek released reasoning models that, by some measures, rival OpenAIs o1-preview. The same month, Chinese videogame juggernaut Tencent unveiled Hunyuan-Large, an open-source model that the companys testing found outperformed top open-source models developed in the U.S. across several benchmarks.holds its own against top performing closed systems from OpenAI and Anthropic.Before DeepSeek-v3 was released, the trend had already caught the attention of Eric Schmidt, Googles former CEO and one of the most influential voices on U.S. AI policy. In May 2024, Schmidt had confidently asserted that the U.S. maintained a two-to-three year lead in AI, which is an eternity in my books. Yet by November, in a talk at the Harvard Kennedy School, Schmidt had changed his tune. He cited the advances from Alibaba, and Tencent as evidence that China was closing the gap. This is shocking to me, he said. I thought the restrictions we placed on chips would keep them back.Beyond a source of national prestige, who leads on AI will likely have ramifications for the global balance of power. If AI agents can automate large parts of the workforce, they may provide a boost to nations economies. And future systems, capable of directing weapons or hacking adversaries, could provide a decisive military advantage. As nations caught between the two superpowers are forced to choose between Chinese or American AI systems, artificial intelligence could emerge as a powerful tool for global influence. Chinas rapid advances raise questions about whether U.S. export controls on semiconductors will be enough to maintain America's edge.Building more powerful AI depends on three essential ingredients: data, innovative algorithms, and raw computing power, or compute. Training data for large language models like GPT-4o is typically scrapped from the internet, meaning its available for developers across the world. Similarly, algorithms, or new ideas for how to improve AI systems, move across borders with ease, as new techniques are often shared in academic papers. Even if they werent, China has a wealth of AI talent, producing more top AI researchers than the U.S. By contrast, advanced chips are incredibly hard to make, and unlike algorithms or data, they are a physical good that can be stopped at the border.The supply chain for advanced semiconductors is dominated by America and its allies. U.S. companies Nvidia and AMD have an effective duopoly on datacenter-GPUs used for AI. Their designs are so intricatewith transistors measured in single-digit nanometersthat currently, only the Taiwanese company TSMC manufactures these top-of-the-line chips. To do so, TSMC relies on multi-million dollar machines that only Dutch company ASML can build.The U.S. has sought to leverage this to its advantage. In 2022, the Biden administration introduced export controls, laws that prevent the sale of cutting-edge chips to China. The move followed a series of measures that began under Trumps first administration, which sought to curb Chinas access to chip-making technologies. These efforts have not only restricted the flow of advanced chips into China, but hampered the countrys domestic chip industry. Chinas chips lag years behind, U.S. Secretary of Commerce Gina Raimondo told 60 minutes in April.Yet, the 2022 export controls encountered their first hurdle before being announced, as developers in China reportedly stockpiled soon-to-be restricted chips. DeepSeek, the Chinese developer behind an AI reasoning model called R1, which rivals OpenAIs O1-preview, assembled a cluster of 10,000 soon-to-be-banned Nvidia A100 GPUs a year before export controls were introduced. Smuggling might also have undermined the export controls effectiveness. In October, Reuters reported that restricted TSMC chips were found on a product made by Chinese company Huawei. Chinese companies have also reportedly acquired restricted chips using shell companies outside China. Others have skirted export controls by renting GPU access from offshore cloud providers. In December, The Wall Street Journal reported that the U.S. is preparing new measures that would limit Chinas ability to access chips through other countries.Read more: Has AI Progress Really Slowed Down?While U.S. export controls curtail Chinas access to the most cutting-edge semiconductors, they still allow the sale of less powerful chips. Deciding which chips should and should not be allowed has proved challenging. In 2022, Nvidia tweaked the design of its flagship chip to create a version for the Chinese market that fell within the restrictions thresholds. The chip was still useful for AI development, prompting the U.S. to tighten restrictions in October 2023. We had a year where [China] could just buy chips which are basically as good, says Lennart Heim, a lead on AI and compute at the RAND corporations Technology and Security Policy Center. He says this loophole, coupled with the time for new chips to find their way into AI developers infrastructure, is why we are yet to see the export controls have a full impact on Chinas AI development.It remains to be seen whether the current threshold strikes the right balance. In November, Tencent released a language model called Hunyuan-Large that outperforms Metas most powerful variant of Llama 3.1 in several benchmarks. While benchmarks are an imperfect measure for comparing AI models overall intelligence, Hunyuan-Larges performance is impressive because it was trained using the less powerful, unrestricted Nvidia H20 GPUs, according to research by the Berkeley Risk and Security Lab. They're clearly getting much better use out of the hardware because of better software, says Ritwik Gupta, the author of the research, who also advises the Department of Defenses Defense Innovation Unit. Rival Chinese labs DeepSeek-v3, believed to be the strongest open model available, was also trained using surprisingly little compute. Although there is significant uncertainty about how President-elect Donald Trump will approach AI policy, several experts told TIME in November that they expected export controls to persistand even be expanded.Before new restrictions were introduced in December, Chinese companies once again stockpiled soon-to-be-blocked chips.This entire strategy needs to be rethought, Gupta says. Stop playing whack-a-mole with these hardware chips. He suggests that instead of trying to slow down development of large language models by restricting access to chips, the U.S. should concentrate on preventing the development of military AI systems, which he says often need less computing power to train. Though he acknowledges that restrictions on other parts of the chip supply chainlike ASMLs machines used for manufacturing chipshave been pivotal in slowing Chinas domestic chip industry.Heim says that over the last year, the U.S.s lead has shrunk, though he notes that while China may now match the U.S.s best open source models, these lag roughly one year behind the top closed models. He adds that the closing gap does not necessarily mean export controls are failing. Lets move away from this binary of export controls working or not working, he says, adding that it may take longer for China to feel them bite.The last decade has seen a dizzying increase in the compute used for training AI models. For example, OpenAIs GPT-4, released in 2023, is estimated to have been trained using roughly 10,000 times more compute than GPT-2, released in 2019. There are indications that trend is set to continue, as American companies like X and Amazon build massive supercomputers with hundreds of thousands of GPUs, far exceeding the computing power used to train today's leading AI models. If it does, Heim predicts that U.S. chip export restrictions will hamper China's ability to keep pace in AI development. Export controls mostly hit you on quantity, Heim says, adding that even if some restricted chips find their way into the hands of Chinese developers, by reducing the number, export controls make it harder to train and deploy models at scale. I do expect export controls to generally hit harder over time, as long as compute stays as important, he says.Within Washington, right now, there is a hesitation to bring China to the [negotiating] table, says Scott Singer, a visiting scholar in the Technology and International Affairs Program at the Carnegie Endowment for International Peace. The implicit reasoning: [If the U.S. is ahead], why would we share anything?But he notes there are compelling reasons to negotiate with China on AI. China does not have to be leading to be a source of catastrophic risk, he says, adding its continued progress in spite of compute restrictions means it could one day produce AI with dangerous capabilities. "If China is much closer, consider what types of conversations you want to have with them around ensuring both sides' systems remain secure, Singer says.

What just happened? There's always an absolute boatload of new tech products on show at CES, some of which are more useful than others. Your opinion of Corsair's Xeneon Edge touchscreen might depend on how much you desire a 14.5-inch, 32:9 secondary display that's also touchscreen. Secondary displays are far from new, of course, but they've become increasingly complex in modern times. The Xeneon Edge is certainly fancy, or a bit weird, boasting a 2,560 by 720 LCD touchscreen, which offers a very crisp 183 PPI density, 350 nits of brightness, and a 60Hz refresh rate on the IPS panel.The Xeneon Edge is flexible when it comes to where you put it. Users can place the device on their desk using the included stand, attach it to a PC or any ferromagnetic surface via the 14 built-in magnets, or put it inside a case at a 360mm radiator mounting point, which does sound pretty cool, admittedly. Corsair writes that it's thinner than a regular case fan, too, so clearance should be of minimal concern.The display connects via USB Type-C DP-Alt Mode or a standard HDMI port. Something else that sounds quite interesting is its ability to operate in both vertical and horizontal orientations. That means it will be useful to anyone who has the need to constantly scroll through their social media feeds or keep an eye on Discord. Windows will detect the Xeneon Edge as another screen, so you can use it as such.Corsair writes that the Xeneon Edge has a five-point multi-touch capacitive touchscreen that functions as a standard Windows touchscreen display for added versatility.The press image shows the display running several apps, including a clock, calendar, weather, and a music player, on the screen simultaneously. It's unclear if this is software that will come bundled with the device. // Related StoriesThe screen also integrates with Corsair's iCue software to show comprehensive information about a system's fan speeds, temperatures, CPU and GPU utilization. You can also alter profiles, lighting, fans, etc., via the touchscreen.No word yet on how much the Xeneon Edge will cost, but don't be surprised if it's quite pricey. The screen will be available in Q2 2025 from Corsair resellers and the company's website.We've seen plenty of LCD screens on the pump caps of AIO liquid coolers before. There was this Full HD one from Lamptron last year that also doubled as a secondary screen, and last March saw Chinese hardware manufacturer Tryx unveil the world's first AIO coolers to feature a curved AMOLED display.

The big picture: TSMC's state-of-the-art $65 billion chip factory in Arizona is reportedly churning out some of the most sought-after processors on the market already. Sources indicate that the facility, known as Fab 21, began manufacturing chips for AMD and Apple earlier this year. Tim Culpan, a well-connected industry insider, has revealed that Fab 21 is already producing at least three major chip designs as part of its initial installed capacity.Last September, it was confirmed that Apple's A16 Bionic SoC for the iPhone 15 and iPhone 15 Plus would be manufactured at the Arizona facility. According to Culpan's report, the fab is now also producing chips for the next-generation Apple Watch.While the exact model remains unclear, Culpan says he is "99% sure" it's the S9 processor package. This specialized system-in-package design is expected to integrate the main application processor core alongside silicon dedicated to AI and neural processing tasks.The biggest surprise from TSMC Arizona is its production of AMD's new Ryzen 9000 series CPUs. Codenamed "Grand Rapids," according to sources, these processors utilize TSMC's cutting-edge 4nm-class N4 and N4P process technologies and form a key part of AMD's latest mainstream lineup. However, the report did not specify which Ryzen 9000 models are currently in production.With three major product lines now being built at the Arizona facility, TSMC's $65 billion investment appears to be paying off. Culpan reports that the fab is currently operating at a capacity of approximately 10,000 wafer starts per month during its initial phase (Phase 1A). That number is projected to more than double to 24,000 wafer starts as part of Phase 1B, though the expansion is reportedly facing a "bottleneck" due to "tooling delays." // Related StoriesOne of the fab's most significant challenges has been staffing and worker retention. While TSMC has prioritized hiring locally now employing more local workers than foreign transfers it has still relied heavily on expertise from Taiwan. This week, the company issued an internal call for hundreds of additional experienced workers from its Taiwan headquarters to help fill critical roles in fab operations and equipment installation.Despite these hurdles, TSMC appears determined to keep its ramp-up schedule on track. In a noteworthy move underscoring the strategic importance of the Arizona facility, the company's board of directors is set to hold a meeting on-site. This rare occurrence will likely include a full tour of the fab, allowing board members to witness the manufacturing process firsthand.