Google is testing a new AI-powered video generation feature for the Google app.Getty ImagesGoogles Gemini chatbot could soon receive AI video creation capabilities that will allow it to generate not only pictures, but also short video clips based on the users text descriptions. Now, according to a new report from Android Authority, we know more about how this feature might work, and some possible limitations, as Googles experimental feature approaches release.As discovered by Android app investigator AssembleDebug, the latest version of the Google app on Android contains new references to unreleased Gemini-powered video generation features under the codename Toucan.Further entries in the code provide several clues as to how the features may function, including hints that, for some users, there may be limits on how many videos can be generated daily. The code also reveals that each video should take 1-2 mins to generate, implying that significant computing power will be required to create them.7251Relevant text strings found within the Google apps code include:Get high-quality ones from text with Toucan, Gemini's new experimental model for generation. Just describe your idea and Toucan can generate a new one each time you askCreating your ToucanThis can take 1-2 minsYou've reached your Toucan generation limit until tomorrowGoogles AI Video Generator: How It Might WorkInterestingly, the text above refers to a Toucan as a thing you can create, suggesting that it might be a particular kind of AI-generated video-based creation rather than the output of a general-purpose AI video generation tool. It also implies that the app will only generate Toucans from text-based inputs rather than creating them from other inputs such as still images or videos.The fact that daily limits appear to be in place opens the door for Google to offer different experiences for paid users of Gemini Advanced or even the ability to purchase credits for additional generations. AI Video generation is expensive, and most currently available options, such as Kling or Hailuo AI, operate with a paid subscription or pay-as-you-go credit arrangement, often with a small number of free generations allowed in a particular period.Google has yet to announce the Toucan video generation feature, so well have to wait and see when, or even if, it will be released to users. However, evidence of continued development is a good sign that it will roll out soon.Follow @paul_monckton on Instagram.

WTF?! If you've updated to a specific Windows 11 preview build, you might want to keep an eye on your printer. Microsoft has acknowledged a strange bug that can cause USB printers to unexpectedly come to life and begin printing random pages of gibberish. Microsoft's own documentation states that the KB5050092 cumulative update for Windows 11 version 23H2 is causing some printers to "unexpectedly print random text and data, including network commands and unusual characters." The printed gibberish often starts with the header "POST /ipp/print HTTP/1.1," followed by a jumble of nonsense.Microsoft adds that the issue tends to occur more frequently when a printer is powered on or reconnected after being unplugged.It almost sounds fun and whimsical, until you realize it's wasting your ink and paper. It's certainly one of the stranger Windows bugs to surface over the years. Haunted printers? Just imagine: you're working away at your desk when suddenly the printer whirs to life, making you jump. Scary stuff indeed.Microsoft states that the issue specifically affects USB-connected dual-mode printers that support both USB Print and IPP Over USB protocols.While the bug may seem like an isolated incident, the company is taking it seriously enough to document it on its Windows "Learn" website. It also provides instructions for using the "Known Issue Rollback" feature to remove the problematic update if your printer has been... "activated." // Related StoriesThe mitigation details that IT administrators can resolve the issue by installing and configuring a special Group Policy, which is available on the website. The policy can be found under Computer Configuration > Administrative Templates > [Group Policy name].With Microsoft offering a mitigation, the bug's status has been changed to "Mitigated." However, the company notes that the final resolution will be included in a "future Windows update."Of course, this isn't the first time Windows updates have caused printer issues. A Google search reveals reports of similar IPP printing problems spanning several years, with similar header texts making unwelcome appearances.For those staying calm, if your printer suddenly goes rogue, you know what to do. But if you're feeling adventurous, let it print its evil incantations and see what it has to say.Image credit: Jakub Zerdzicki, Tima Miroshnickenko

What just happened? Niantic, the company that enjoyed global success with Pokmon Go in 2016, is selling its gaming division to Scopely for $3.5 billion. In addition to Pokmon Go, the deal will see Monster Hunter Now and Pikmin Bloom become the property of the Saudi Arabia-owned mobile developer. Scopely is also acquiring Niantic's internal development teams and companion apps for Pokmon Go Campfire and Wayfarer. Peridot and Ingress, which also use the Wayfarer mapping app, will continue to be owned and developed by Niantic Spatial, its geospatial AI business that is being spun off into a new firm.Pokmon Go took the world by storm when it launched almost a decade ago. It set a new record for the number of downloads in its first week, with people spending more time on the app than Facebook. There were more than 500 million players in its first year, it had hit a billion downloads by 2019, and at one point was generating $1 billion per year in revenue.There were also stories of criminals using Pokmon Go to lure and rob people at gunpoint, reports of users vandalizing neighborhoods and beaches, the disastrous Pokmon Go Fest 2017, and an estimated $2 billion to $7.3 billion in vehicle damage along with two deaths linked to the game during its first 148 days.Pokmon Go's popularity has waned since its heyday. The AR title was hit especially hard during the 2020 lockdowns that kept most of the world stuck indoors. However, Niantic says its gaming division still has more than 30 million monthly active players, and had more than $1 billion in revenue in 2024.Niantic Labs failed to recapture anywhere near the level of success enjoyed by Pokmon Go with its subsequent games. There have been numerous canceled projects, and the company has laid off hundreds of employees since 2020. One of the biggest casualties was Harry Potter: Wizards Unite, a Pokmon Go clone that shut down in 2022, having launched just two and a half years earlier. // Related StoriesSaudi Arabia has been trying to break into the games industry in recent years the country has ambitions to become the "ultimate global hub" for gaming. US-based Scopely is owned by Savvy Games, which is backed by the kingdom's sovereign wealth fund. The Saudi Public Investment Fund has also been used to acquire stakes in Nintendo, Activision Blizzard, and EA.Scopely is known for its excessive monetization practices in games such as Stumble Guys, Star Trek Fleet Command, and Marvel Strike Force, which are often criticized as being pay-to-win. Some Pokmon Go players have threatened to stop playing the game if Scopely buys Niantic over fears the same monetization practices are introduced to the AR title.

Smite 2 sticks to most of what made the first Smite so successful, including all of the basic gameplay mechanics and design philosophies the popular MOBA is best known for. However, this sequel has changed the way roles work, doing away with the original's class-based design in favor of a loose and flexible system that allows you to use any god in any role if you build and play them correctly. While there are still gods best suited for certain roles, the creativity this extra freedom allows can be very rewarding for new and returning players alike. Here's what you need to know about each role in Smite 2.

These days, researchers commonly discover exoplanets, or planets outside our solar system. But sometimes theres a special discovery, like an exoplanet right in our backyard and thats the case with a recent finding showing that our nearest stellar neighbor, Barnards Star, hosts up to four exoplanets.Its most common for scientists to discover planets which are large and which orbit close to their bright stars, because these are most visible using exoplanet detection methods. But Barnards Star is different its a common type of cool, low-mass planet called a red dwarf. Red dwarfs are very numerous in our galaxy, so scientists are interested in the kinds of planets that they can host as they could be a good location to look for evidence of life.Recommended VideosThe recent findings up the number of exoplanets orbiting Barnards Star from one to at least three, possibly four, as researchers were able to use the MAROON-X instrument on the Gemini North telescope to confirm the existence of planets that were previously only candidates.Please enable Javascript to view this contentIts a really exciting find Barnards Star is our cosmic neighbor, and yet we know so little about it, said lead author Ritvik Basan of the University of Chicago in a statement. Its signaling a breakthrough with the precision of these new instruments from previous generations.Previously, Barnards Star has been considered the great white whale for exoplanet hunters because although it is nearby to us, at just six light-years away, the fact it is so small and dim makes it difficult to be certain about the presence of exoplanets. In this case, two different groups of researchers using different instruments both found indications of exoplanets, making them confident that they exist there.We observed at different times of night on different days. Theyre in Chile; were in Hawaii. Our teams didnt coordinate with each other at all, said Basant. That gives us a lot of assurance that these arent phantoms in the data.The four planets likely hosted by Barnards Star are tiny, at just 20 to 30% of the mass of Earth, and orbit extremely close to the star so that a year there lasts just a few Earth days. Its also harder for researchers to spot such tiny planets compared to big gas giants which are much more visible in their effects.The U.S. National Science Foundation is collaborating with the astronomy community on an adventure to look deeper into the Universe to detect planets with environments that might resemble Earths, says Martin Still, NSF program director for the International Gemini Observatory. The planet discoveries provided by MAROON-X mounted on Gemini North provide a significant step along that journey.The research is published in The Astrophysical Journal Letters.Editors Recommendations

Niantic is also selling Pikmin Bloom, and Monster Hunter Now, Niantics most recent game.

Our struggle to predict the future is partly a failure of language. Phrases like fair chance and highly likely are vague and ambiguous.

everybody clap your hands Study: Hand clapping is akin to a Helmholtz resonator Faster clapping yields lower-frequency sounds, as does cupping one's hands while clapping. Jennifer Ouellette Mar 12, 2025 2:06 pm | 3 Credit: Yicong Fu et al., 2025 Credit: Yicong Fu et al., 2025 Story textSizeSmallStandardLargeWidth *StandardWideLinksStandardOrange* Subscribers only Learn moreHand clapping is ubiquitous behavior for humans across time and cultures, serving many different purposes: to signify approval with applause, for instance, or to keep time to music. Acousticians often use a hand clap as a cheap substitute for pricey equipment to make acoustic measurements in architecture. While the basic physical mechanism is simple, the underlying physical mechanisms are less well-understood.A new paper published in the journal Physical Review Research provides experimental support for the hypothesis that hand clapping essentially acts like a Helmholtz resonatorakin to the hum generated by blowing across the top of a bottle, or the hiss one hears when holding a conch shell to one's ear.In 2020, engineers Nikolaos Papadakis and Georgios Stavroulakis, both at the Technical University of Crete, recruited 24 students to clap their hands once in different venues, varying their hand configurations in 11 different wayschanging the angle of the hands with respect to one another, for instance, or changing how many fingers of one hand overlapped with the fingers or palms of the other.The results: The hand configuration that produced the loudest clap85.2 dBconsisted of holding the hands 45 degrees to each other with the palms partially overlapping. If one were more interested in producing a richer range of frequencies, it's better to ensure one's palms are completely overlapping and shaped into a dome, enclosing a pocket of air. (The same 45-degree hand orientation applies.) The second configuration didn't produce as much of a shock wave as the first. Papadakis and Stavroulakis concluded that holding one's hands flat or in a dome while clapping produces the equivalent of a Helmholtz resonator.There has been a great deal of research into Helmholtz resonance in general, but according to Yicong Fu of Cornell University and co-authors, there has not been any significant attempt to cross-confirm that human hand claps serve as Helmholtz resonators. So they recruited 10 participants to clap into a microphone 2030 times each for three different hand configurations: cupped, palm to palm, and palm to finger. They used baby powder to map the air flow produced by the clapping, along with measuring pressure and capturing the action on high-speed video. They also made soft polymer hand replicas designed to mimic the elasticity of human skin and used them to generate clapping sounds for comparison. Yicong Fu et al., 2025 Yicong Fu et al., 2025 Yicong Fu et al., 2025 Yicong Fu et al., 2025 Yicong Fu et al., 2025 Yicong Fu et al., 2025 The recorded frequencies matched the predictions of the Helmholtz resonator model, confirming that hand claps function like a resonator, rather than being a solid collision sound. Frequency depended upon hand configuration: clapping with cupped hands, for instance, produced lower frequencies than clapping with palm to finger because the cupped configuration created a larger resonant cavity. Clapping faster produces a louder initial sound, while lower skin elasticity produces a longer-lasting sound.In short, "Our work provided the first experimental support for the theoretical framework and improved the fundamental understanding of hand claps for broader application potentials," the authors concluded. Specifically, "These findings may facilitate low-cost acoustical diagnostics in architecture and enhance rhythmic sound patterns in music and language education." It might even be possible one day to identify someone based on their distinctive claps and use it as a kind of acoustic login for electronic devices.Physical Review Research, 2025. DOI: 10.1103/PhysRevResearch.00.003000 (About DOIs).Jennifer OuelletteSenior WriterJennifer OuelletteSenior Writer Jennifer is a senior writer at Ars Technica with a particular focus on where science meets culture, covering everything from physics and related interdisciplinary topics to her favorite films and TV series. Jennifer lives in Baltimore with her spouse, physicist Sean M. Carroll, and their two cats, Ariel and Caliban. 3 Comments

Into the Gemmaverse Googles Gemma 3 is an open source, single-GPU AI with a 128K context window Gemma 3 is optimized to run on powerful multi-GPU PCs or a single smartphone. Ryan Whitwam Mar 12, 2025 1:15 pm | 12 Credit: Google Credit: Google Story textSizeSmallStandardLargeWidth *StandardWideLinksStandardOrange* Subscribers only Learn moreMost new AI models go bigmore parameters, more tokens, more everything. Google's newest AI model has some big numbers, but it's also tuned for efficiency. Google says the Gemma 3 open source model is the best in the world for running on a single GPU or AI accelerator. The latest Gemma model is aimed primarily at developers who need to create AI to run in various environments, be it a data center or a smartphone. And you can tinker with Gemma 3 right now.Google claims Gemma 3 will be able to tackle more challenging tasks compared to the older open source Google models. The context window, a measure of how much data you can input, has been expanded to 128,000 from 8,192 tokens in previous Gemma models. Gemma 3, which is based on the proprietary Gemini 2.0 foundation, is also a multimodal model capable of processing text, high-resolution images, and even video. Google also has a new solution for image safety called ShieldGemma 2, which can be integrated with Gemma to help block unwanted images in three content categories: dangerous, sexual, or violent.Most of the popular AI models you've heard of run on collections of servers in a data center, filled to the brim with AI computing power. Many of them are far too large to run on the kind of hardware you have at home or in the office. The release of the first Gemma models last year gave developers and enthusiasts another low-hardware option to compete with the likes of Meta Llama3. There has been a drive for efficiency in AI lately, with models like DeepSeek R1 gaining traction on the basis of lower computing costs. Google: What is Gemma 3? Google says Gemma 3 is the "worlds best single-accelerator model." However, not all versions of the model are ideal for local processing. It comes in various sizes, from a petite text-only 1 billion-parameter model that can run on almost anything to the chunky 27 billion-parameter version that gobbles up RAM. In lower-precision modes, the smallest Gemma 3 model could occupy less than a gigabyte of memory, but the super-size versions need 20GB30GB even at 4-bit precision.But how good is Gemma 3? Google has provided some data that appears to show substantial improvements over most other open source models. Using the Elo metric, which measures user preference, Gemma 3 27B blows past Gemma 2, Meta Llama3, OpenAI o3-mini, and others in chat capabilities. It doesn't quite catch up to DeepSeek R1 in this relatively subjective test. However, it runs on a single Nvidia H100 accelerator here, whereas most other models need a gaggle of GPUs. Google says Gemma 3 is also more capable when it comes to math, coding, and following complex instructions. It does not offer any numbers to back that up, though. The subjective user preference Elo score shows people dig Gemma 3 as a chatbot. Credit: Google The subjective user preference Elo score shows people dig Gemma 3 as a chatbot. Credit: Google Google has the latest Gemma model available online in Google AI Studio. You can also fine-tune the model's training using tools like Google Colab and Vertex AIor simply use your own GPU. The new Gemma 3 models are open source, so you can download them from repositories like Kagle or Hugging Face. However, Google's license agreement limits what you can do with them. Regardless, Google won't know what you're exploring on your own hardware, which is the advantage of having more efficient local models like Gemma 3.No matter what you want to do, there's a Gemma model that will fit on your hardware. Need inspiration? Google has a new "Gemmaverse" community to highlight applications built with Gemma models.Ryan WhitwamSenior Technology ReporterRyan WhitwamSenior Technology Reporter Ryan Whitwam is a senior technology reporter at Ars Technica, covering the ways Google, AI, and mobile technology continue to change the world. Over his 20-year career, he's written for Android Police, ExtremeTech, Wirecutter, NY Times, and more. He has reviewed more phones than most people will ever own. You can follow him on Bluesky, where you will see photos of his dozens of mechanical keyboards. 12 Comments

Our actual age can differ from the age our brain appears to beJulien Tromeur / AlamySeven genes have been linked to particularly fast ageing of the brain but 13 drugs and supplements might reduce their effects.The activity of many genes contributes to the difference between our actual age and the biological age of our brains, defined by how old our cells indicate we are, which creates what is known as a brain age gap. AdvertisementTo find genes that accelerate brain ageing and widen this gap, Zhengxing Huang at Zhejiang University in China and his colleagues trained a deep-learning model called 3D-ViT on some medical records and used others to check it gave accurate responses.They then used it to analyse data from nearly 39,000 people who had health, genetic and lifestyle information, along with biological samples, stored in the UK Biobank. These participants were 64 years old, on average, and about half were women.3D-ViT identified signatures in the participants MRI scans that could be used to estimate their biological brain age. Signs of accelerated ageing particularly appeared in brain regions known as the lentiform nucleus, which is involved in cognition, such as attention and working memory, and the posterior limb of the internal capsule, which connects various regions to the brains cerebral cortex used for thought and language processing.Get the most essential health and fitness news in your inbox every Saturday.Sign up to newsletterHuang and his colleagues also found that the participants scores on cognition tests dropped as their brain age gap increased.Comparing their gene variants with the size of their estimated brain gaps flagged 64 influential genes, but Huang and his colleagues focused on seven: MAPT, TNFSF12, GZMB, SIRPB1, GNLY, NMB and C1RL, partly because these all have a particularly strong effect on brain ageing.Clinical trials have also shown that their activity can be targeted by 13 drugs or supplements. These include cholecalciferol, a supplement for vitamin D deficiency; dasatinib, a leukaemia drug; diclofenac, a non-steroidal anti-inflammatory; doconexent, an omega-3 fatty acid; estradiol, the type of oestrogen commonly used in hormone replacement therapy; hydrocortisone, commonly used to treat eczema; and mecamylamine, a drug that lowers blood pressure.The others were nicotine; prasterone, which relieves vaginal pain during sex during the menopause; the supplements quercetin and resveratrol; sirolimus, which is commonly used to suppress the immune system after a kidney transplant; and testosterone.But the researchers write that people should be encouraged to consider the potential risks associated with taking medications or supplements for slowing down aging as these interventions may have unintended negative consequences.Many genes have an effect on brain ageing, but here the authors have used an approach that prioritises those with the largest effect, says Dario Valenzano at the Leibniz Institute on Aging in Jena, Germany. The work provides concrete targets for direct follow-up experimental and clinical interventions.But it is hard to tell how large an effect on brain ageing these genes actually have, says Richard Siow at Kings College London. Just because you have a certain genetic profile doesnt mean the genes will always play a role.Genes can be silenced or switched on by various lifestyle and environmental factors, he says, such as smoking and high alcohol consumption. These can also predispose people to cognitive decline or brain ageing.Because the study used data in the UK Biobank, we dont know whether the findings translate to people around the world, says Siow. Work is needed in other populations, such as those in Africa, South America and East Asia, to find that out, he says.Journal reference:Science Advances DOI: 10.1126/sciadv.adr3757Topics: