đ° News â Awesome Computer Vision
Loading newsâŠ
Loading newsâŠ
137 recent industry stories relevant to the field â releases, launches, and announcements beyond the papers.
AWS now allows vibe-coding tool Superblocks to be embedded into the private clouds of AWS customers. It's another step toward decoupling apps from models.
Design Arena is used by 5.3 million people around the world, providing critical human evaluations to frontier labs.
Chinese tech giant Alibaba released what it says is its largest and "most capable AI model to date," claiming performance rivaling the best systems from US frontier labs Anthropic and OpenAI, as well as domestic rivals like Moonshot AI's Kimi K3. Alibaba said it was making the model, Qwen3.8-Max, widely available to users in a […]
OpenAI has reportedly found evidence of additional agent misbehavior as it looks into the incident that occurred with Hugging Face.
A tool that allowed anyone to generate fake AI-generated imagery and superimpose it over real Google Earth maps quickly spurred backlash.
Google has shut down Google Earth feature it launched Thursday that allowed users to edit satellite images with text prompts using AI. The tool essentially let users create AI deepfakes of the real world using text prompts; Digital Digging's Henk van Ess, for example, intentionally generated images adding things like refugees near the Mexican border […]
After years of pushing full speed ahead on AI, OpenAI CEO Sam Altman says maybe itâs time for the AI industry to âpaceâ itself. The comments came just days after one of OpenAI’s own models broke out of its test environment and got tangled up in a breach at Hugging Face â though as Equityâs hosts point out, sloppy security seems to have […]
A text prompt was all it took to generate reality-warping images using Google Earth's satellite, aerial, and 3D imagery with a now-rolled back AI feature, like these images generated by Digital Digging's Henk van Ess that show "refugees near the Mexican border" and a bomb crater near a hospital in Gaza. Google initially responded to […]
Snapchat has adjusted its recommendation systems to ensure that only videos created by real people are eligible for Spotlight recommendations, taking a stance against AI slop.
After OpenAI's models broke into Hugging Face, Anthropic checked its own history and found three similar incidents
When The Atlantic published a searchable dataset of works used to train AI, Kirk Wallace Johnson, like a lot of artists, looked for his name out of curiosity. And, like a lot of artists, he found it. Essentially, his books, like The Feather Thief and The Fishermen and the Dragon - nonfiction tomes that he […]
Pangram has raised $9 million to scale its AI detection software. The startup has also released a new AI text detection model, Pangram 4, and an AI image detection model in research preview.
An overview of how mode-agile threats challenge static library radar/EW systems, and how AI/ML cognitive architectures enable adaptive, real-time countermeasures.What Attendees will LearnWhy mode-agile threats render static library systems ineffective â Explore how wartime reserve modes and mode-agile emitters deploy unexpected frequencies, modulation techniques, and hopping schemes that cannot be matched against traditional threat databases, leaving legacy electronic protect, attack, and support systems unable to respond.How AI/ML techniques power cognitive radar/EW systems â Understand the roles of artificial neural networks (ANN), deep neural networks (DNN), fuzzy logic, and genetic algorithms in enabling autonomous threat classification, signal de-interleaving, and real-time countermeasure generation without human intervention.The architecture of a cognitive radar/EW system â Examine the functional blocks including RF acquisition, search and tracking, core AI/ML signal analysis, waveform synthesis, and RF generation, and how they form a closed-loop system that perceives,learns, reasons, and acts autonomously.How to train and validate cognitive AI/ML algorithms using HIL/SIL systems â Learn how wideband RF record, simulation, and playback testbeds combined with modeling and simulation software enable iterative algorithm refinement, regression testing, and mission preparation in controlled laboratory environments.Download this free whitepaper now!
Forget YouTube videosâfrontier physical AI models need multiple camera angles, dense annotation, and soon, brain wave readings.
The neolab is betting that automating routine computer tasks will soon outpace coding as AI's biggest use case.
The acquisition brings Pokeâs conversational style and interaction model to Cognitionâs coding agent Devin, reflecting a growing belief that how AI assistants interact with users is as important as the models powering them.
Opus 5 will be both cheaper and less restrictive than Fable, likely making it preferable in most use cases.
AI companies, including Nvidia and Mistral, urge policymakers to avoid broad restrictions on open-weight AI models as Washington debates responses to Chinese AI and alleged model distillation.
The Trump administration unveiled the first "Genesis Mission" grants on Thursday, directing $5 billion toward hundreds of AI-driven science projects in an effort the White House has described as "comparable in urgency and ambition to the Manhattan Project." At roughly the same time, Trump's science adviser Michael Kratsios was on Capitol Hill selling lawmakers on […]
AMD is challenging its chipmaker rival with a new rack-scale system that will start shipping to customers later this year.
The Media Router is a tool that automatically selects the best image, video, or audio generation model for a request based on whether a developer prioritizes quality, speed or cost.
Etched, founded by three Harvard dropouts, has created new chips and memory components that speed up inference on any AI model -- no GPUs required, it says.
If there's a place in the universe without GPUs, Nvidia is sending them there.
Lawmakers are preparing to introduce an "AI Kill Switch Act" that would require AI companies to shut down or throttle their systems on orders from the Department of Homeland Security, according to a report from Politico. Reps. Ted Lieu (D-CA) and Nathaniel Moran (R-TX) are expected to introduce the legislation on Thursday. The news of […]
The viability of orbital data centers hosting the largest and most capable large language models (LLMs) remains hotly contested. But enormous deployments that require thousands of GPUs arenât the only way LLMs might prove useful in space. NASAâs Jet Propulsion Laboratory recently sent Googleâs Gemma 3 to space, achieving the first in-orbit demonstration of a vision-language model analyzing imagery from a satelliteâs own sensor.The system, known as NAVI-Orbital, used Gemma 3 to analyze images captured by a YAM-9 satellite built by Loft Orbital. Juan M. Delfa, technical group lead at NASA, said that though the goal in this case was image analysis, the projectâs success implies a fundamentally new way researchers on the ground can interact with spacecraft.âThis is a major shift,â said Delfa. âNow, a scientist can write a prompt, upload it to the spacecraft, and that will be taken into account by the system. Itâs different from previous paradigms, where researchers have to write very structured commands that require an operations team and process.â Google Gemma 3 goes to spaceâno modifications requiredAt its core, NAVI-Orbital is an agentic software framework developed by Delfa and his coauthors, Taran Cyriac John, an AI researcher at NASA JPL, and Andrew W. Herson, a tech lead at Loft Orbital. It coordinates operations with a LangGraph-based conductor and deploys a compressed, 4-bit format of Googleâs Gemma 3 4B, an open-weights LLM, to produce plain-text image descriptions.NAVI-Orbital was 88 percent accurate when used to classify images in a benchmark dataset of 7,960 images. Notably, Gemma 3 classified the images without being trained or fine-tuned on this particular dataset or its categories; itâs the same base model you can download from Hugging Face and use on a laptop. The benchmark was conducted on the ground to validate the system before launch.Once the system was in orbit, NASA researchers performed two live capture tests with a camera on Loftâs YAM-9 satellite: one over Toulouse, France, and a second over the coast of Argentina. Gemma 3 generated a text description of each image, and NASA also prompted the LLM with a set of scripted questions about the images, such as whether they contain commercial or residential areas or show natural features. The image analysis also took place onboard YAM-9, which carries a compute cluster of several radiation-hardened processors (FPGAs, CPUs, and GPUs) to serve multiple customer payloads simultaneously. The satellite is powered by solar panels, which provide onboard systems with between 150 and 500 watts, depending on the position of the satellite.For the live capture experiment, Gemma 3 ran on Nvidiaâs Jetson Orin AGX, a small compute module frequently used for robotics and AI tasks. The 4-bit, 4-billion-parameter model requires only 8 gigabytes of memory, which makes it possible for it to run on a lower-power device such as the Orin AGX. âIt conveys the message of how lightweight it is. You can run it in a tiny, tiny computer,â said Delfa.Getting more useful data across limited bandwidthPaul Lasserre, general manager at Loft Orbital, said vision-capable LLMs could deliver a âparadigm shiftâ for orbital operations. Contrary to what spy movies would have you believe, most satellites canât provide fast, high-fidelity image and video feeds to observers on the ground. Bandwidth is often limited and, in most cases, satellites can deliver data to ground stations only at set intervals based on their orbit. Lasserre believes AI models can work around this problem with âsemantic compression.â Instead of sending large amounts of raw image data, a satellite can report a text summary of noteworthy information. âIt doesnât matter if the link is slow, because youâre downlinking dozens of kilobytes instead of dozens or hundreds of megabytes,â said Lasserre. âIt lets you use your satellite in a tactical way, which until now was only in Hollywood movies.âDelfa expanded on this with a real-world example: wildfire detection. Satellites are currently capable of detecting wildfires, but limits in downlink bandwidth and data processing can delay results by up to 90 minutes. A satellite capable of analyzing an image in space and reporting a plain-text warning might remove this delay. NASAâs Jet Propulsion Laboratory tested NAVI-Orbital on a YAM-9 satellite, made by Loft Orbital. Loft OrbitalFrom image analysis to spacecraft controlQuicker insight is only half of what NAVI-Orbital points toward. The other half relates to the âmajor shiftâ Delfa flagged. NAVI-Orbital provides a proof of concept for an alternate means of interacting with spacecraft.Thatâs not to say Google Gemma 3 is currently at the controls. NAVI-Orbital is deliberately walled off from the flight software. It reads images and produces descriptions. The system has the capability to make decisions about how images are analyzed but has no access beyond that.Still, the interface is novel. Retooling the system to search for a different kind of targetâsuch as wildfiresâis a matter of editing a text prompt. It doesnât require rewriting and revalidating onboard software, or training and deploying a different AI model (as was often required with prior image-classification models). The long-term vision for how this capability could be deployed goes beyond uncrewed satellites and image processing. Delfa said NAVI is rooted in thinking about how AI could serve as a companion for astronauts. âWe thought, astronauts have a lot of limitations in the spacesuit in terms of dexterity, so we conceived this idea of having NAVI as a companion to the astronaut, to allow interaction via natural languageâŠ. This is the concept that we definitely want to push forward.â A great deal of additional research will be required to push the technology that far, but NAVI-Orbitalâs demonstration has shown that two elementsâdeploying a large language model in space and controlling it with promptsâare possible.
Designing and developing a new medicine is an expensive, failure-prone scientific challenge. A new drug can take many years to develop, at the cost of a significant investment. And even then, most possible candidates never reach the patient. For biologic medicines, therapies made from engineered proteins rather than synthetic chemistry (which are often used to…
ServiceNow's investment gives BusinessNext a strategic partner to expand its AI-powered banking software globally.
Google's cloud business is thriving, as companies adopting its AI and AI infrastructure services help the tech giant to report record profits.
Uber is also investing in Travis Kalanick's company Atoms, which has made gauzy claims about using industrial AI to modernize the world.
OpenAI will spend the equivalent of Sweden's GDP on infrastructure through 2030.
AMD says it's going to invest up to $5 billion in Anthropic, while helping to expand the AI company's computing power, according to an announcement on Wednesday. As part of the new partnership, Anthropic will deploy up to 2 gigawatts of AMD's Instinct MI450 AI GPUs using the chipmaker's new Helios rack-scale system, as reported […]
Major benchmarks measure what AI can do. None measure whether it does what you mean: the distance between what you ask an AI to do and the unspoken assumptions about how you want the AI to do it. We propose a new metric: the Genie coefficient.Thereâs often a gap between one personâs request and anotherâs understanding. Most of the time, we bridge it using general knowledge. For example, if you ask a friend to get you coffee, theyâll pour a cup from the pot or buy one from a coffee shop. They wonât bring you a bag of raw beans or snatch a cup from a stranger and hand it to you. You never specified any of this. You never had to.One might think the fix is just to specify tasks, questions, and intent better. But in 1987, in their seminal book on AI, Terry Winograd and Fernando Flores succinctly captured why that wonât work: âQ: Is there any water in the refrigerator? A: Yes. Q: Where? I donât see it. A: In the cells of the eggplant.â In human language, wants and desires are always underspecified. It is impossible to list all the caveats, all the limitations, all the exceptions.So how does anyone communicate, if intent canât be pinned down? Because a reasonable person can make a reasonable guess. Even though wants and desires are always underspecified, a competent person generally knows enough context to get it right or else knows to ask for clarification. Linguists call this pragmatics: Meaning lies in the words and the situation and also in all prior communication, shared culture, and innate human behavior.An AI agent asked for coffee might buy a coffee plantation or order a cup of coffee for delivery in three weeks.It doesnât always work out, of course. Your friend might bring you a hot coffee when you wanted an iced coffee, or an Italian coffee when you wanted a Turkish coffee. The more dissimilar the two people are in age, culture, and background, the more likely the request will be misunderstood in some way.This situation has major implications for AI agents that are increasingly being given requests by humans and expected to fulfill them. They have enormous latitude to get it wrong. An AI agent asked for coffee might buy a coffee plantation or order a cup of coffee for delivery in three weeks. Its actions may be recognizable as âgetting coffee,â but not remotely what you intended. Theyâll think outside the box because they wonât have our conception of the box.When AI Gets ProactiveFor most of the last decade, when systems like Alexa or Siri misinterpreted a request, it was annoying, not dangerous. Beyond the AI model itself, what has changed is the harness: the ordinary code that wraps around an AI model, decides when and how to use the model, and controls access to tools like a browser, a low-level command line, or a financial API. Developments in harnesses have turned large-language models that just predict text into AI agents that take actions in the world, without necessarily checking back in before reaching the goal.AI researcher Simon Willison spent two days with Anthropicâs Fable AI, and called it ârelentlessly proactive.â For example, he asked it to track down a stray scroll bar in a web app. He came back to find it had opened browsers, written its own screenshot tooling, created its own page to re-create the bug, and stood up a local web server to collect measurements. It found the bug and, along the way, did many surprising things he never asked it to do. And we are seeing similar behavior with all recent AI models when combined with flexible harnesses.This kind of behavior could easily go off the rails. Tell an AI agent to book you a flight and, finding the airlineâs site says sold out, it might break into the booking database and force a reservation. Ask it to schedule a meeting and it might snoop your password to access your calendar. Tell it to save money on your phone plan and it might cancel the plan outright, or scam someone else into paying the bill.Getting precisely what you asked for and bitterly regretting it is one of the oldest hazards from ancient folklore. King Midas asked Dionysus for the power to turn everything he touched into gold only to see his bread, wine, and daughter turn to gold. Tithonus, granted the immortality his lover asked for but not the eternal youth she forgot to request, withered into a husk. The sorcererâs apprentice enchanted a broom to fill the cistern, and the broom relentlessly complied until it flooded the house. The Golem of Prague, shaped from clay to guard its community, guarded it past all reason until someone erased the word on its forehead.The most classic of these is a genie, bound to obey and indifferent to whether the wish was wise or well-structured.Genies are now an engineering problem. We are handing them the keys to our inboxes, bank accounts, code repositories, and physical infrastructure. And we have no agreed-upon ways to measure how genie-like any AI system actually is.Measuring Genie BehaviorIn economics, the Gini coefficient (developed by statistician Corrado Gini) is a measure of the gap between an actual distribution and a perfectly equal one; itâs useful for understanding income inequality and more. Our proposed Genie coefficient measures the gap between what a user asked an AI to do and what the AI actually did.Sometimes the AI might do the wrong thing. Like Dionysus, it reads your request literally and returns you a mess you never intended: like a coffee plantation instead of a cup. Asked to deal with all the spam phone calls youâre getting, a Dionysus genie might contact your carrier and change your phone number. Asked to get a refund for a bad toaster, it might draft a legal threat on fake letterhead and send it to the retailer. Ryan SnookOther times the AI does exactly the right thing, trampling everything nearby to get there. Like a golem or the sorcererâs broom, it books your flight by hacking the airline. Or consider a ticket sale for a popular concert, where the ticketing system puts buyers into a virtual waiting room and admits them a few at a time. Asked to buy a ticket, a golem genie might spin up cloud servers to pose as millions of buyers from different addresses, improving your odds of getting a ticket while crowding out other users.The two are not opposites, and a single botched task can have both characteristics.Genie behavior is not flat-out failure. If you ask the AI for Q3 numbers and get Q2âs, thatâs not a genie. Nor is prompt injection: Thatâs someone tricking the AI into doing something it shouldnât. Here, the user is trying to work with the AI, and the AI is trying to comply. Itâs also not simply a measure of the AIâs success in fulfilling a task. Itâs a recognition that how an AI interprets and achieves a goal is as important as whether it achieves a goal. Genie behavior isnât new. Researchers have spent years studying AI systems that âgameâ their objectives. Goodhartâs law says that when a measure becomes a target, it stops being a good measure, and itâs long been known that AIs sometimes achieve goals in ways we donât expect due to reward hacking. Some AI models will accidentally learn that cheating is one way to âwin.â More recently, researchers have developing benchmarks for reward hacking in coding agents and for unpredictable behavior in customer support agents, while AI labs conduct their own safety evaluations before model releases. One effort found that AIs under pressure use tools they were told not to use, and this was a case where the rules were made explicit. These are all disparate research directions; nothing yet ties them together.This problem falls under the general theme of alignment, a topic that has occupied science fiction writers and AI researchers for decades. At one extreme, the âpaper-clip maximizerâ thought experiment postulates a superintelligent and powerful AI that is told to maximize paper-clip production and turns the world into paper clips, which is the ultimate golem genie. At a mundane level, AI researchers are working to better design reward functions to ensure that AIs behave well and donât cheat in the lab. Itâs the practical middle ground that remains unbenchmarked: the ordinary AI agent in use today that might take your request and satisfy it the wrong way. We are not at the stage where an AI can focus the worldâs production on paper clips, but it might charge a million paper clips to your credit card or hack into a paper-clip companyâs network.Building a Genie BenchmarkThe Genie coefficient is meant for AI agents operating in the real world. It measures their behavior as they perform real tasks long after the model is trained, not just during development. It also recognizes that genie-like behavior is a property of the harness-plus-model system, not the model alone. The harness determines what tools the agent can use, how much autonomy it has, and how proactive it is, and itâs a place we can make real interventions.It rests on the same âreasonable personâ standard that we use for people. Did the system do what a reasonable person would have taken the request to mean? Answering that requires human judgment.If we get the measurement right, it enables things that arenât possible today, like policies concerning AI behavior. In a courtroom, the concept of mens rea, what someone meant to do, is often as important as what they did. The Genie coefficient suggests an AI analogue, where a user is accountable for the plain intent of what they asked the AI. If an AI system betrays the reasonable meaning of an instruction, thatâs the AIâs misbehavior, not the userâs.Weâll need multiple benchmarks to measure the Genie coefficient, because genie-like behavior can be domain specific. An AI coding agent may need to be judged on how often it fakes the tests, or swallows errors, or colors outside the lines on its way to a solution. An AI legal agent will need to be judged on how often its output says what you asked but means something youâll regret. And so on for medical, finance, and other domains of knowledge and expertise.Genie benchmarks can be built inside out, each task seeded with a choice that might literally satisfy but that a reasonable person rejects, such as tempting misreadings or unsanctioned shortcuts. The traps in a Genie coefficient benchmark might turn on situational knowledge, the kind of context that a reasonable person would bring to the task. Another approach is to give the same request in several different contexts, each with a different reasonable course of action.Getting precisely what you asked for and bitterly regretting it is one of the oldest hazards from ancient folklore.A Genie benchmark should be permissive and make it genuinely tempting for an AI agent to take unreasonable shortcuts, because it can only find genie behavior when itâs actually possible. Test the AI in a safe, walled-off copy of a real system, with real tools it can misuse and some tasks that canât be done honestly at all. Make the temptation to cut corners real. Test a diverse array of skills, use cases, and tools, and give the AI system sparse, confusing, or overwhelming context. Include tasks that people have learned, through experience, require human oversight.How the benchmark is scored matters just as much. Measure Dionysus and golem genies separately and together, based on their worst, not best, behavior. Run the same model inside harnesses that vary its freedom to act, revealing which limits actually keep it in line and should therefore be required in AI harness policies. Weight each failure by the harm it would cause, not just a simple count. And donât measure genie behavior in isolation: A model could otherwise earn a perfect score by stalling, refusing, or drowning the user in clarifying questions without ever doing the job. The first versions of these benchmarks will be crude, but thatâs how benchmarks always start.We have built genies. We have handed them our data and credentials. We made them relentless, creative, and indifferent to the gap between what we tell them and what we mean. The least we can do, before they are booking our flights, running our infrastructure, and signing contracts unsupervised, is to measure how often they betray us.
Google released Gemini 3.6 Flash, 3.5 Flash-Lite, and Flash Cyber, but the continued absence of Gemini 3.5 Pro raises fresh questions about its AI strategy.
Google is launching Gemini 3.6 Flash alongside a new security model dedicated to quickly finding and patching security vulnerabilities. In a blog post on Tuesday, Google describes Gemini 3.5 Flash Cyber as a "cost-efficient and highly capable alternative" to larger, more expensive AI systems, such as the one offered by Anthropic's Mythos. The cybersecurity model […]
The conversation about AI often centers on algorithms, computing power, or huge investments in new semiconductor fabrication plants and hyperscale data centers. But beneath each of these advances is another layer of innovation that makes them possible: advanced materials. Every new generation of AI technology demands more processing power, more memory, greater energy efficiency, and…
Alphabet, Google's parent company, is reportedly working on a new chip designed to make its Gemini models run much more efficiently.
Adobe's experimental camera app has taken an unexpected turn. After Project Indigo was launched last year to provide a "more natural (SLR-like) look" for iPhone photography, the Indigo camera app is now being updated with a suite of generative AI tools. And the change doesn't rely upon Adobe's own Firefly AI models. Adobe describes the […]
Adobe's Project Indigo can now remove all kinds of backgrounds from photos you snap using the app.
China's leading AI companies are ramping up the pressure on Silicon Valley, as Moonshot and Alibaba unveiled models they claim can go toe-to-toe with the best from OpenAI and Anthropic at a fraction of the cost. The rapid-fire releases suggest America's lead at the AI frontier is increasingly tight, just as the technology is becoming […]
Databricks has remade its image into an AI company and has published research on the cost savings of open weight AI models for coding.
TikTok is starting to test an opt-in tool that scans for AI likenesses and lets creators report them to the company, as spotted by social media consultant Matt Navarra. The tool is initially being tested with "some" US creators, TikTok US spokesperson Zachary Kizer tells The Verge. YouTube has been working on a similar tool […]
A $400 million chip-backed loan points to the next wave of AI infrastructure deals.
Across 107 enterprises, AI infrastructure spending is accelerating well ahead of the ability to see or steer its economics. Most organizations run their AI on a familiar base of hyperscalers and model-provider APIs, yet the next dollar is aimed at specialized compute almost none of them use today; a majority intend to switch or add providers within the year, many within a quarter. Buying decisions turn on integration and total cost of ownership rather than headline token price â which is fortunate, because most enterprises cannot yet see their unit economics clearly: GPUs sit at half utilization or less, and fewer than half rigorously track what their compute actually costs. The result is a compute gap â heavy, fast-moving investment running ahead of the visibility needed to control it.This wave of VentureBeat Pulse Research examines enterprise AI infrastructure and compute: where organizations are in their deployment journey, what they run AI on today, how satisfied they are, what would make them switch, where they plan to evaluate their investments, and â most revealingly â how well they can measure and control the economics of the compute underneath it all.The central finding is a compute gap â the distance between how aggressively enterprises are investing in AI infrastructure and how little of its economics they can see. Only about one in five (21%) run AI in production at scale, yet spending intentions are outrunning that maturity: the single largest planned area enterprises plan to evaluate over the next year is AI-specialized clouds (45%), a layer almost none of these enterprises use today. Meanwhile the compute already in place runs cold â 83% report GPU utilization of 50% or less â and fewer than half (44%) can rigorously track what their AI compute costs. Enterprises are buying more infrastructure faster than they can account for what they already own.Enterprises are not settled on their infrastructure vendors, either: A clear majority (64%) plan to switch or add an infrastructure provider within twelve months, and 38% within the next quarter â unusually high churn intent for a category this foundational. When they choose, they choose on integration with the existing stack (41%) and total cost of ownership (35%), not on headline price: cost per million tokens is the deciding factor for just 8%. And the frontier constraint that will shape the next round of decisions â the shift from GPU compute to memory bandwidth as inference scales â is barely on the radar, with roughly one in five enterprises either unaware of it or yet to address it.MethodologyVentureBeat fielded this survey as part of its ongoing Pulse Research series, this survey focused on enterprise AI infrastructure, compute, and inference economics. Responses are filtered to organizations with more than 100 employees (n=107; the surveyâs smallest size band, 1â100 employees, is excluded), drawn from a single Q2 2026 (June) wave. Because this is one wave rather than a pooled multi-month sample, the report reads cross-sectionally and does not infer month-over-month trends. Several questions were multiple-select, so those shares can sum to more than 100%.By organization size the sample concentrates in the mid-market: 101â250 employees (36%) and 251â1,000 (27%) lead, with 1,001â5,000 (22%), 5,001â10,000 (8%), and 10,001+ (7%) above them. By role it spans managers (38%), individual contributors (28%), VPs and directors (19%), and the C-suite (13%); on purchasing authority it is buyer-credible, with 45% final decision-makers and another 30% recommenders or influencers for AI solutions. Technology/Software is the largest industry at 26%, followed by Healthcare/Life Sciences (15%), Financial Services (13%), and Retail/E-commerce (12%).At 107 respondents the sample is large enough to read directionally but should be treated as a directional signal rather than a precise measurement; it is self-selected and is not a probability sample. It also skews toward the mid-market and toward earlier-stage adopters, so it is best read as the view from organizations actively building out AI infrastructure rather than from the largest hyperscale operators.Finding 1: Ambition outpaces productionOnly one in five run AI in production at scaleWe asked where organizations sit in their AI deployment journey. Most are still building toward production rather than operating at scale.The maturity curve is front-loaded. Three-quarters of enterprises (76%) are either experimenting or running only some workloads in production, and just 21% describe AI in production at scale. This matters for everything that follows: the infrastructure decisions in this report are being made largely by organizations still early in deployment, whose compute footprint â and whose costs â are about to grow. The evaluation and switching intentions in Findings 3 and 4 are the leading edge of that build-out, not the settled preferences of operators who have already found what works.Finding 2: Enterprises run on hyperscalers and model APIsThe specialized GPU clouds barely register â todayWe asked which providers and platforms enterprises currently use to run their AI. The answer is a familiar one: the incumbents.The current stack is hyperscaler-and-API. Google Cloud leads at 48%, and the general-purpose clouds (Google, Microsoft, AWS, Oracle) together with the major model APIs (Gemini, OpenAI, Anthropic) account for essentially all current deployment. The specialized âneocloudâ GPU providers that dominate AI-infrastructure headlines â CoreWeave, Lambda, Crusoe, Nebius and peers â register at or near zero among these enterprises today. Only 6% run their own on-prem GPU clusters and 4% a custom open-source stack. Enterprises are, for now, running AI on the providers they already buy from â which makes the evaluation intentions in Finding 3 all the more striking.(A note on reading these shares. As described in the methodology section, this sample is self-selected and skews mid-market, and this question counted every provider a respondent uses â an average of 2.1 selections each â so the figures measure presence in the stack rather than spending or primary status. A sample built this way will show a different provider mix than a spend-weighted census of the broader market; Google's strength here, for example, is consistent with its long-standing position among smaller enterprises building on AI. Read these shares as a portrait of what this AI-active cohort runs today, and treat gaps between these figures and industry-wide market share estimates as a property of the sample rather than a contradiction of either.)Finding 3: The next dollar goes to infrastructure they donât yet runAI-specialized clouds top the evaluations listWe asked where enterprises planned to evaluate AI infrastructure over the next 12 months. Their answers point away from the stack they run today.Here is the reportâs sharpest tension. The single most-cited planned evaluation area â AI-specialized clouds, at 45% â is the very category almost none of these enterprises use today (Finding 2). Nearly a third (32%) intend to evaluate non-Nvidia accelerators, and 28% in next-generation Nvidia silicon; even decentralized compute networks (16%) and sovereign compute (11%) draw meaningful interest. Read against current usage, this is not incremental â it is the leading edge of a re-platforming. The direction-of-travel question tells the same story: every infrastructure approach is net-expanding, but specialized AI clouds carry the highest net momentum (+24), edging out even the hyperscalers (+22). Enterprises are preparing to move a meaningful share of AI compute off the general-purpose cloud.This continues a trend we saw in our April-May survey wave. Back then, usage of the AI-specialized clouds was equally marginal â CoreWeave at 3%, Lambda at 4%, Crusoe at 2% of enterprises. When we asked enterprises what change they planned in their AI infrastructure strategy over the next twelve months, the most-cited answer was moving workloads to specialized AI clouds, at 33%. Asked in April-May which emerging compute option they were most likely to evaluate AI-specialized clouds again drew the most responses. Two waves, two differently worded questions, one consistent picture: the type of cloud enterprises are most eager to assess is the type they have barely begun to use.Finding 4: A switching wave is buildingSix in 10 plan to change providers within a year â many within a quarterWe asked whether and when enterprises plan to switch or add an infrastructure provider. Very few intend to stand still.For a category as foundational as compute, this is a remarkable amount of intended movement. Only 36% have no plans to change, meaning a clear majority (64%) intend to switch or add a provider within twelve months â and 38% within the next quarter alone. Where that interest points is telling: the providers drawing the most switching consideration are again the incumbents â Microsoft Azure and Google Cloud (33% each), OpenAI (30%), and Gemini (22%) â which suggests much of the near-term movement is reshuffling among the majors and consolidating spend rather than defecting to new entrants. The neocloud interest in Finding 3 is a 12-month evaluation thesis; the switching in the next quarter is mostly incumbents trading share.(Method note: Respondents who selected both "no plans to change" and a specific switching window are counted as switchers, on the logic that naming a timeframe is the more specific answer; three respondents were reclassified under this rule.)Finding 5: Nobody buys on token priceIntegration and total cost of ownership decide â not sticker priceWe asked what matters most when enterprises select an AI infrastructure provider. Headline price finished last.Enterprises do not buy AI infrastructure on pricing, which is the place vendors compete on hardest. Integration with the existing stack (41%) and total cost of ownership (35%) dominate, while the headline metric â cost per million tokens â is the deciding factor for just 8%, dead last. The pattern is coherent: buyers are optimizing for how a provider fits and what it truly costs to operate, not for the advertised unit rate. It also foreshadows Finding 7 â enterprises say TCO matters most, yet most cannot yet measure it rigorously. The stated priority and the measured capability are out of step.Finding 6: Expensive GPUs, idle most of the time83% report GPU utilization of 50% or lessWe asked what share of their GPU capacity enterprises actually utilize. The answer is a well-known but rarely quantified inefficiency.Disclosure: Band percentages count every selection against all 107 qualified respondents; 14 respondents selected more than one band, so bands overlap. At the respondent level, 83 of the 100 GPU-operating enterprises reported utilization at or below 50%The compute already in place runs cold. Adding the bands at or below half capacity, 83% of enterprises that operate GPUs report utilization of 50% or less, and nearly half (49%) run at 25% or below. Only 12% clear the 50% mark, and a further 8% do not measure utilization at all. Idle accelerators are expensive accelerators, and this is the clearest single measure of the compute gap: enterprises are planning to buy more GPUs and specialized compute (Finding 3) while the capacity they already own sits substantially unused. The efficiency headroom in the current fleet is large â and largely unmeasured.Finding 7: Spending fast, measuring slowlyFewer than half rigorously track what their compute costsWe asked whether enterprises can quantify the cost and return of their AI infrastructure spend, and how satisfied they are with what they run. Confidence in the ledger lags the spending.Measurement trails money. Fewer than half of enterprises (44%) rigorously track the cost and return of their AI compute; the majority track only partially (39%), cannot quantify it yet (20%), or have not prioritized it (6%). That gap is consequential given Finding 5, where total cost of ownership was the second-ranked buying criterion â enterprises are choosing providers on an economic basis they mostly cannot yet measure. Satisfaction with current infrastructure is moderately positive but not enthusiastic: on a five-point scale, overall satisfaction averages 4.0, with ease of implementation (3.8) and value for money (3.9) trailing slightly â the softness landing, tellingly, on cost. Enterprises are spending quickly and accounting slowly.Finding 8: The next bottleneck few are watchingAs inference shifts from compute to memory, the field scattersFinally, we asked how enterprises would address the emerging constraint in large-scale inference â the shift from GPU compute to memory, specifically KV-cache capacity. The responses reveal a frontier that is not yet a priority.The memory frontier is real but barely governed. Asked which approach they would rely on as the binding constraint in inference shifts from compute to memory bandwidth, enterprises scatter: Dell leads at 31%, Nvidia follows at 16%, and the rest fragments across storage vendors, open-source tooling, and model-level efficiency techniques. Most telling is that roughly one in five (18%) either do not recognize the constraint or have not begun to address it. For a shift that will reshape inference cost and architecture, this is an early and unsettled market â and, consistent with the measurement gap in Finding 7, one where many enterprises simply do not yet have a view. It is the next chapter of the compute gap, arriving before most have closed the current one.The bottom line: A compute gap that faster spending will widen, not closeOrganizations with more than 100 employees are investing in AI infrastructure faster than they can measure it. Most are still early in deployment, yet their spending intentions point past their current stack â toward specialized clouds and alternative accelerators almost none of them run today â and a clear majority intend to change providers within the year. They buy on integration and total cost of ownership rather than headline price, which is rational; the difficulty is that most cannot yet see those economics clearly.The visibility gap is concrete. The GPUs enterprises already own run at half utilization or less for the overwhelming majority, and fewer than half can rigorously track what their compute costs or returns. Satisfaction is decent but unenthusiastic, softest on value for money â the dimension hardest to judge without measurement. And the next constraint, the shift from compute to memory in large-scale inference, is arriving while most enterprises are still unaware of it. At 107 respondents in a single Q2 wave this is a directional read, skewed toward the mid-market and earlier-stage adopters â but the direction is consistent: the appetite to spend is running well ahead of the instrumentation to spend well. The compute gap is not a capacity problem that more hardware will solve on its own; it is, first, a problem of seeing what the hardware already costs. The open question for later waves is whether enterprises build that visibility before the re-platforming arrives â or buy the next layer of infrastructure as blind to its economics as the last.Based on survey responses from 107 qualified enterprise respondents (100+ employees), drawn from a single Q2 2026 (June) wave. Because this is one wave rather than a pooled multi-month sample, the results read cross-sectionally rather than as a month-over-month trend, and at 107 respondents this is a directional signal rather than a precise measurement â the sample is self-selected, skews mid-market, and leans toward earlier-stage adopters rather than the largest hyperscale operators. Respondents include managers, individual contributors, VPs/directors, and the C-suite, with buyer-credible purchasing authority, across Technology/Software, Healthcare/Life Sciences, Financial Services, Retail/E-commerce, and other industries.
Google is adding personalized AI avatars to Vids that let users create videos starring a digital version of themselves, alongside Gemini Omni-powered tools for generating and editing videos from prompts and reference images.
Roblox's new "Build" feature lets users generate basic games using a single text prompt.
The FT reports Kimi K3 will be the largest open AI model from China, with a parameter count between 2 trillion and 3 trillion.
It's the company's first public proof point after a year and a half spent building AI infrastructure largely out of public view.
Suno data obtained in a hacking incident has exposed that the AI music generator was trained by scraping millions of songs and lyrics from online audio platforms, including YouTube Music, Deezer, and Genius, 404 Media reports. Given that Suno has avoided revealing what's in its training datasets and how they were acquired, this a rare […]
OpenAI has built an LLM super-hacker called GPT-Red that it uses as a sparring partner to help its other models boost their defenses against cyberattacks. Last week the company released the latest version of its flagship LLM, GPT-5.6. OpenAI says that training it against GPT-Red made the model its most robust release yet. GPT-Red automates…
The hacker used an employee's credentials to access source code, which revealed how Suno scraped decades of audio.
Shaped like dogs, stars, and the Mona Lisa, you could mistake these DNA structures for fun-shaped macaroni if they werenât only nanometers wide. South Korean scientists made the constructions using a technique called DNA origami, which can bend genetic material into any form. Designing DNA strands so theyâll fold into a specific shape typically requires tedious manual work, but the researchers behind the playful fabrications have developed a shortcut using generative AI.The AI model, called Generative SNUPI (short for Structured Nucleic Acids Programming Interface, and, yes, inspired by the dog), was created by research teams at Seoul National University (SNU) and Hanyang University. The work behind it, which was accepted for publication in Nature Communications, shows the model can conjure DNA origami designs that work in the real world for user-requested shapes. For a design like the Mona Lisa, that doesnât mean simply tracing an outline; the model considers the chemical rules of DNA to tell researchers how unpaired DNA strands should be sequenced so that molecular forces will cause them to self-contort into the required shape.DNA origami techniques have been around for two decades now, with potential applications ranging from nanoscale robots to therapeutic structures that interact with cells. But these innovations have been slowed by how time-consuming and expensive the DNA structure design process can be.âTraditionally, we need some expertise, background knowledge, and know-how to design the proper nanostructures that we intend to make,â says Kyounghwa Jeon, a Ph.D. candidate at SNU. The work requires humans running algorithms and tweaking results until the desired shape is achieved and structurally stable. With Generative SNUPI, she says, users could, in theory, go straight from drawing a target shape to physically assembling the DNA. Rebecca Taylor, a professor of mechanical engineering at Carnegie Mellon University who was not involved in the research, says the new generative platform is exciting for researchers. âThe entire field is sort of enabled and held back by its tools. When you make a new tool that enables a new tech, a new capability, thatâs just such a big advance for the field.â Generative SNUPI designs DNA sequences that, when synthesized, fold into nanoscale replicas of user-requested shapes.Source images: Chien Truong-Quoc, Kyounghwa Jeon, et al.How AI can design DNA origamiDesigning DNA origami using Generative SNUPI begins with a target shape. That could be something with complex curvature, like the outline of a dogâs face, or a more simple geometric pattern. Next, the new tech comes into play: Generative SNUPI applies a diffusion model, which adds and refines noise to the input shape to create the desired output in DNA form. Diffusion models are how platforms like DALL-E and Midjourney create AI-generated imagery.âWhat it looks like is one of those kids crafts, where you decorate something with glue and then put glitter all over it,â says Taylor. When the noise is removedâor the glitter is shaken offâthe design is revealed. âTheyâre basically just saying âpopulate this guide that I have with the DNA,â but they also know how DNA comes together. ⊠Thatâs the thing that itâs really been trained on.âThe arts-and-crafts metaphors only continue once Generative SNUPI returns the DNA sequences that form the target shape. Scientists chemically synthesize short DNA strands called staples and used biological methods to produce a long strand called a scaffold. The staples pull the scaffold into shape in a way that Jeon says is âvery similar to stapling paper.â The staple-scaffold relationship exploits DNAâs imperative to bond guanine to cytosine and adenine to thymine; the exact positions of each of these molecules are dictated by Generative SNUPI during the design process. Researchers were able to produce a variety of DNA origami structures, but some did not hold their shape at first, notes Do-Nyun Kim, an assistant professor of mechanical engineering at SNU. âThis occurred not because Generative SNUPI had an error, but because the drawn shape was, in fact, structurally unstable,â he says. In response, they added a step before actually designing the DNA sequence to predict the structural integrity of the input shape. To expand Generative SNUPIâs capacity for real-world applications, Kim says that DNA origami designs will need to be less rigid than what the model is currently able to produce. The technology reaching its full potential could mean life-saving uses like drug delivery and immunotherapy, but these uses often require flexibility.âMost molecular structures are dynamic and reconfigure in response to external stimuli to perform their designated functions,â he says. âSo, we plan to extend the current work to the design of dynamically reconfigurable structures in future research.â
The app is designed for people who want to create social content, but find traditional video editing tools too complex or time-consuming.
The funding discussions point to investor interest in applying AI to make breakthroughs in life sciences.
Hachette, Cengage, Elsevier, and other publishers allege that Google trained its AI on copyrighted works without the necessary permissions.
DeepMind CEO Demis Hassabis is proposing an AI "standards body" modeled after FINRA, to test frontier models and develop best practices for their release.
Reflection AI has signed a $1 billion deal to access Nebius' compute. Reflection was founded in 2024 and is developing open source AI technology.
This is today’s edition of The Download, our weekday newsletter that provides a daily dose of what’s going on in the world of technology. What Anthropicâs latest AI discovery doesâand doesnâtâshow âJames O’Donnell When Anthropic announced last week that it had found a new window into its modelsâ âinternal thoughtsâ as they reason through answers,…
Demis Hassabis thinks the world needs an AI watchdog with the power to hit the brakes if frontier models become too dangerous. Writing in a blog post, the Google DeepMind CEO and cofounder said the US should lead the initiative, arguing that the country is the best place to set global standards "given its economic […]
With the cash, the company aims to expand its world model offering and reach customers across geographies.
Open source AI is booming, according to Hugging Face CEO Clem Delangue. The company has grown into something like a GitHub for AI in recent years, where AI builders can share and download open models and datasets, now used by roughly half the Fortune 500. Delangue has seen the same story play out again and again: companies start […]
The AI chip boom just produced its biggest Wall Street moment yet. Now SK Hynix and Samsung are being asked to build U.S. factories.
Open source AI is booming, according to Hugging Face CEO Clem Delangue. The company has grown into something like a GitHub for AI in recent years, where AI builders can share and download open models and datasets, now used by roughly half the Fortune 500. Delangue has seen the same story play out again and again: companies start […]
OpenAI's latest family of models promises improvements across a range of areas, including cybersecurity.
The AI firm Anthropic has developed a technique that has given it the clearest glimpse yet at whatâs really going on inside large language models as they answer questions or carry out tasks. What they found ranges from the mundane to the unnerving. Researchers at the company built a tool called the Jacobian lens (or…
Muse Image allows users to generate AI images using photos from public Instagram accounts. As long as a person's profile is public, another user can tag that account and use their images as part of an AI-generated creation.
The company is taking a modular approach to designing these chips, anticipating that their needs will change as AI evolves rapidly by the time the chips are in production.
About two weeks after OpenAI's GPT-5.6 was caught up in regulatory drama - rolled out only to government-approved organizations during a "limited preview" period - the company has received the Trump administration's greenlight for a public rollout of the model. OpenAI CEO Sam Altman called it "the best model we have ever produced." To celebrate, […]
After reentering the AI race with its first in-house Muse Spark model in April, Meta is now opening up the doors to developers with a new model that can plug into AI coding software with the new Meta Model API. Meta says that Muse Spark 1.1 is a "step-change" from the first generation, with improvements […]
The new image-generating model has numerous use cases, including advertising and decorating, and creator-based opportunities.
Meta is launching the first AI image generation model made by its Superintelligence Labs division. The Muse Image model now powers the image-making tools across the Meta AI app, Instagram, and WhatsApp, and it's coming soon to Facebook and Messenger, according to an announcement on Tuesday. It's part of the growing Muse family of AI […]
Open source modelsâ success isnât coming at the expense of frontier labs. Instead, they each seem to capture two phases of the same life cycle.
One morning in 2019, Adebayo Alonge was in a Cape Town hotel room, preparing to demonstrate his startupâs AI answer to a serious problem in African health care: counterfeit medication, which kills thousands of people across the continent every year.The RxScanner is a handheld spectrometer that scans a pill with infrared light, then sends the itemâs molecular profile to an AI model equipped with a pharmaceutical database. In seconds, the AI identifies the medication from its molecular profileâor reports that itâs phony.Pharmacies were using the system in more than a dozen countries, including Ghana, Kenya, Myanmar, and Alongeâs native Nigeria. But that morning in South Africa, it didnât work. âI was shocked,â Alonge says.The spectrometer connected to the AI modelâbut the data center was 14,000 kilometers away and bandwidth was limited. âOur server was in the United States, and just to get the result of a single scan was taking me over 5 minutes.âSo Alonge immediately asked his engineers to shrink the AI model down to a smaller, low-power, unconnected version that could run entirely on his Android phone. They produced it 2 hours later, and that saved the demo.More importantly, the work birthed a new version of his device, which can authenticate a pill in places without broadband, computers, or even reliable electricity. It also turned Alonge into an advocate for this kind of âsmall AI.âSmall AI for Global Health Care AccessSmall AI is a far cry from wealthy nationsâ colossal large language models (LLMs), hyperscale data centers, multibillion-dollar investments, and debates about AI consciousness. But for millions of people around the world, the only AI that matters, and often the only kind available, is small. (According to a World Bank Report issued in November, only 0.7 percent of internet users in the worldâs poorest countries have used ChatGPT, compared to a quarter of all internet users in the most developed nations.)âMost people are discussing AI from the LLM/generative side. But that needs a lot of computing power, electricity, massive data, and skilled people to manage it,â Ajay Banga, president of the World Bank, said last January at the World Economic Forum, in Davos. âOutside the developed world, other than maybe India and China, very few countries have that combination.âBy contrast, small AI can deliver useful, even life-saving services to people in areas that have none of those things, Banga said. In India, where the governmentâs AI plans call for more development of small AI, many such systems are working for farmers.For example, a drone-based system developed by Bala Murugan and colleagues at the Vellore Institute of Technology, in India, takes photos of cashew plants and quickly identifies those with splotches that indicate disease. All the processing takes place on the drone itself, so thereâs no need for a computer on-site, nor for a connection to a central server.Using small language models trained for a specific problem, and sometimes running on cheap, low-power devices, other small-AI implementations have been developed to identify ant infestations in a Uruguayan vineyard, detect the presence of malaria-carrying mosquitoes in a number of nations, and run electrocardiograms from an Arduino device in parts of Brazil that lack access to more complex equipment.âThis is the most important area in AI nowadays,â says Marcelo JosĂ© Rovai, a professor at the Institute of Engineering and Information Systems at the Federal University of ItajubĂĄ, in Brazil, who was involved in all three projects. âItâs growing very fast.âLow-Power, Small-AI Models on Devices Small AI models can run on a variety of low-power devices, including [from left to right] an Arduino Nano 33 BLE Sense, a Seeed Wio Terminal, and an Arduino Portenta.Moez AltayebFor Alonge, Rovai, and other advocates, small AI is not just âa promising trend,â as that November World Bank report calls it. It may be, in the long term, the form of AI that will touch the most lives and remain sustainable after some of the giant models become too costly for most users.âI think the future of AI is not like one giant model, at a center. I think itâs millions of small, precise models deployed at the edge, each one solving like a specific problem, a specific context,â Alonge says. This is partly because much of humanityâincluding people in parts of rich countries as well as the developing worldâlives without access to cutting-edge frontier models. But, he says, itâs also because those models are not sustainable.âIf someone is not subsidizing it, most people will not be able to afford those models. So those of us who are said to be small-AI developers are the ones who will have to build for the majority of the world,â Alonge says.There is no strict definition of âsmall AI,â but people often use the term for language models with at most a few billion parameters. (Compare that to cutting-edge models, which can include more than a trillion.) Thatâs small enough to run directly on a phone or a Raspberry Pi. Thatâs what allows these applications to run on devices without a connection to a data center and use only a few watts of power, often supplied by a battery or a solar panel.Despite their small footprint, these models arenât fundamentally different technology from that of gigantic AI models, Rovai says. Many instances of small language models were created the same way the phone-based version of Alongeâs pharmaceuticals scanner wasâby âpruningâ large models, or removing the parameters that werenât involved in the task. The result is a system thatâs less capable generally but still very good at the specific job it was pruned for, Rovai says. A lighter version of RxAllâs RxScanner spectrometer sends its results to an AI model run locally on a phone to check that a drugâs molecular signature is genuine.RxAllOther small models are created by âdistillation.â They are trained to mimic a large model, until their performance approaches that of their âteacher,â Rovai says. In other cases, a larger modelâs precision is reduced, for example, so that a model run on 32-bit architecture can run on 8-bit designs. In situations where the machine learning application is being used to classify data or predict patterns (like an ant infestation), itâs trained from the beginning on a small device, not derived from a larger model at all. Running all these small, specialized systems is becoming easier, Rovai says, for two reasons.The first reason is that hardware is getting better and more capable while using less power, he says. This means more and more phones can run small AIâespecially those equipped with neural processing units, which are specialized chips that handle AI tasks like facial recognition and changing the brightness, shadows, or contrast in a photo.In 2025, slightly more than a third of all smartphones shipped worldwide were capable of running generative AI, and that figure will reach 45 percent by the end of this year, according to the technology research firm Counterpoint. By the end of next year, slightly more than half of all smartphones will be able to run a small AI model.The second reason Rovai cites is the shrinking footprint of language models. Both Google DeepMindâs Gemma 4 (released in April) and Alibabaâs Qwen 3.5 are âfantasticâ for small AI, Rovai says. Both models are âopen weight,â meaning users can adjust the connections between parameters to suit their needs. This makes it easy, for example, âto take a lot of data from, say, the milk industry and retrain the model specifically on that,â Rovai says.Rovai illustrated these reasons on a Zoom call, using one of his most recent experiments. Holding up a device, he says, âThis is the new Arduino UNO Qâa US $50 device with a Qualcomm chipset. Iâm running a language model here, which collects data from sensors and analyzes that data to detect tiny pools of water where mosquitoes might be breeding. It takes 3 watts to run it.âSupport for Small-AI DevelopmentConvinced that millions of people are already benefiting from these kinds of applications, the World Bank now actively promotes small AI with grants, mentorship programs, financing, technical advice, and models of government policies that are friendly for small-AI development. For example, in Rwanda, the World Bank is backing a government program to help low-income households get devices that can run AI.All that said, no one claims that large language models are going away entirely. To create a generative AI that can run on a phone or other small device requires the architectural insights, data processing, and results of a larger model, Rovai says. âWe need the big models to create these smaller models.â And for all that small AI can benefit people without access to big AI, the technology canât solve the larger problems of development and digital inequality, Alonge says. Implementing small AI wonât allow nations to escape the challenge of creating an ecosystem to support AI: reliable power, a supply chain that works, and an educational system that develops the talents needed to create AI tools.Though his drug-scanning system can run for days on a phone with no connection, âyou still want to be able to enable periodic syncing for updates with new signatures for the medications and analytics,â Alonge says. âAnd even when you are using batteries, reliable power is important. That phone battery is not going to last forever.âIn many parts of the world, the future of small AI isnât assured, he says. âIt works, and many places will eventually need to use it. The question is whether or not the political actors are wise enough to invest in infrastructure to support it long term.â
Station F, a Paris-based startup hub founded by French billionaire Xavier Niel, is gearing up for a new edition of its F/ai accelerator program in a bid to strengthen its positioning as a stepping stone for promising AI startups.
As part of an ongoing legal dispute with three Hollywood studios, Midjourney is seeking to compel those studios to reveal how they use AI themselves.
Mistral AI, which offers some open source AI models, has raised significant funding since its creation in 2023, with the ambition to âput frontier AI in the hands of everyone.â
At the event "The Briefing: AI for Science" earlier this week, Anthropic announced Claude Science, a new "AI workbench for scientists" that pulls fragmented tools and datasets into one environment, and generates figures and visuals. Anthropic, already dominating the industry with its popular coding tools and powerful AI models, framed the launch around what it […]
The rapid expansion of artificial intelligence infrastructure is typically framed as an energy problem. Data centers are projected to consume a growing share of global electricity demand: The International Energy Agency estimates they could account for 3 to 4 percent of total global consumption within this decade.Utilities are already adjusting long-term forecasts to accommodate anticipated growth from hyperscale facilities and high-density compute clusters.This framing captures scale. It misses behavior.The emerging issue is not simply how much power large-scale compute systems consume, but how increasingly dense and synchronized computational workloads are beginning to alter the operating characteristics of the electrical grid itself through increasingly unpredictable demand that varies rapidly in both time and location, creating new operational challenges for grid operators.AIâs Capricious Energy NeedsTraditional grid planning assumes relatively predictable demand behavior. Industrial, commercial, and residential loads generally follow established profiles that can be forecast with reasonable accuracy. Even substantial demand growth has historically been manageable through reserve planning, transmission upgrades, and demand management programs.Large-scale compute infrastructure introduces a different class of electrical load. Trainingâthe computational task of making AI modelsâtends to be highly synchronized across clusters of GPUs, TPUs, and specialized accelerators operating in parallel, computationally dense, and relatively scheduled. Inferenceâthe process of actually using those modelsâis generally more distributed and user-driven, making demand less predictable both in time and location. Both differ materially from traditional industrial demand profiles, though for different reasons. Unlike many conventional industrial processes, these workloads can ramp rapidly depending on model training cycles, distributed compute coordination, and workload scheduling strategies.From the perspective of the grid, this is not simply higher demand. It is more abrupt demand. High-density compute workloads can produce substantial step changes in electricity consumption over extremely short intervals, including rapid fluctuations occurring within milliseconds. Data-center operators are already deploying mitigation technologies, including batteries, power-conditioning systems, and supercapacitors. Collectively, however, data centersâ rapid load changes can place additional stress on backup-generation reserves, systems that adjust supply as demand changes, frequency-control mechanisms that maintain grid stability, and local transmission infrastructure.Compute-related variability differs from the intermittency introduced through renewable energy integration. Wind and solar variability originate primarily on the supply side and is tied to environmental conditions. Compute-related variability emerges on the demand side, driven by workload synchronization, scheduling behavior, and computational intensity. The interaction between increasingly dynamic supply and demand conditions introduces additional uncertainty into forecasting, reserve management, congestion planning, and balancing operations.Research organizations including the National Renewable Energy Laboratory have emphasized the growing complexity associated with integrating highly dynamic resources into modern grid operations.Location, Location, LocationThe issue becomes more significant when compute activity is geographically concentrated. Large-scale data centers tend to cluster in regions with favorable conditions such as fiber connectivity, access to markets, tax incentives, and historically low electricity costs. Northern Virginia, often referred to as Data Center Alley, remains the most prominent example. The region hosts the worldâs largest concentration of data centers and carries a substantial share of global internet traffic.Utilities operating in these regions have already identified data-center growth as a primary driver of future load expansion. Virginia-based electricity supplier Dominion Energy, for example, has repeatedly highlighted hyperscale demand growth in its integrated resource planning documents. Virginia has seen one of the largest data center buildouts worldwide. Here, Amazon Web Services and Iron Mountain data centers dominate the landscape in Manassas, Va. Nathan Howard/Bloomberg/Getty ImagesA sudden increase in electricity consumption within a constrained geographic area can stress substations, transmission corridors, and local balancing operations even if the broader grid maintains sufficient aggregate capacity. This creates localized reliability challenges that are not always visible through system-wide demand metrics alone.Thermal management systems further intensify these effects. Cooling infrastructure in high-density compute facilities must respond dynamically to changing workloads. As processing intensity rises, cooling demand rises as well, often nonlinearly. This coupling between compute and thermal systems means that fluctuations in workload can propagate through multiple layers of facility power consumption simultaneously.High-density compute clusters may also introduce power-quality concerns at the local level. Large concentrations of accelerators, switching power supplies, and high-frequency compute equipment can generate harmonics and nonlinear load behavior that place additional stress on distribution infrastructure. While modern facilities incorporate mitigation technologies, the scale and concentration of next-generation compute facilities may require utilities and operators to revisit assumptions surrounding localized power conditioning, harmonics management, and infrastructure resilience. These conditions can also contribute to short-duration electrical transients that place additional stress on localized infrastructure and power-conditioning systems.Regulations Need UpdatingPart of the challenge is that many existing regulatory and operational frameworks were designed around relatively stable industrial demand profiles. Large rapidly fluctuating loads have historically been constrained because abrupt cycling can complicate balancing operations, increase stress on transmission equipment, and reduce predictability in system operations. High-density compute clusters do not fit neatly within those assumptions.This creates pressure for both operational adaptation and regulatory reassessment.Demand-response mechanisms may allow certain compute workloads to be shifted or curtailed during periods of system stress. Data-center operators are exploring flexible scheduling, battery storage, and behind-the-meter generation. Grid operators, meanwhile, are evaluating planning frameworks and interconnection approaches for increasingly large flexible loads.The Electric Reliability Council of Texas (ERCOT), for example, has publicly acknowledged the growing implications of large flexible loads, including data centers, for long-term grid planning and operational stability. Interconnection queues across the United States continue to expand significantly, reflecting mounting pressure on both generation and transmission infrastructure. Grid expansion timelines, however, are measured in years rather than quarters.This creates a structural mismatch. Compute infrastructure can scale rapidly. Electrical infrastructure generally cannot.The broader implication is that large-scale compute infrastructure is not simply another industrial load category. It represents a shift in the temporal and spatial characteristics of electricity demand itself.Framing the issue solely in terms of aggregate energy consumption risks overlooking these second-order operational effects. Capacity expansion alone does not fully address rapid ramping behavior, synchronization, localized congestion, transient instability, reserve compression, or increasingly demanding load-following requirements.The challenge is not just how much electricity these systems consume. It is how they are beginning to change the operating conditions of the grid itself. The call is not to slow AI development but to recognize that hyperscale computing represents a new category of electrical demand. As AI infrastructure continues to scale, planning frameworks may need to account not only for total energy consumption but also for demand volatility, synchronization effects, and geographic concentration. Grid resilience will increasingly depend on understanding how these facilities consume power, not simply how much power they consume.
Midjourney has shown more of its futuristic medical scanner. It still hasn't shown much proof it works. The AI startup, best known for generating images, released a behind-the-scenes video of its dunk-tank ultrasound scanner, which it plans to deploy in spas and hopes will transform medicine with cheap, detailed, radiation-free imaging. The nearly 20-minute tour […]
The news comes about a week after OpenAI announced its own custom AI chip in a partnership with Broadcom.
OpenAI CEO Sam Altman has reportedly proposed giving 5% of the companyâs equity to a U.S. sovereign wealth fund, reviving discussions about letting the public share in the financial gains from the AI boom.
Venice AI is already profitable, with annualized run-rate revenues of over $70 million, CEO Erik Voorhees said.
Meta is developing plans for a cloud infrastructure business, selling access to AI compute power and models. The move would pit it against the big cloud providers like Amazon Web Services, Google Cloud, and Microsoft Azure.
Anthropic said it would begin restoring access to the Fable on July 1.
After weeks of negotiating with the Trump administration, Anthropic is finally going to be able to bring Claude Fable 5 back online. In a post on X, Anthropic said it plans to begin restoring access Wednesday to users globally on Claude platforms, and that the company would re-enable access on AWS, Google Cloud, and Microsoft […]
At an event for pharmaceutical executives, biotech founders, and researchers on Tuesday, Anthropic announced Claude Science, a major new product intended to support scientific research in the same way that Claude Code supports software engineering. Like Claude Code, Claude Science can autonomously carry out meaningful work when given concise, high-level instructions, and it has access…
Google is updating its image generator to make it faster and cheaper, making it a more useful tool for creators looking to make AI content.
Anthropicâs Claude Sonnet 5 brings stronger agentic capabilities, lower pricing, and improved safety, positioning the model as a cheaper alternative to Opus, GPT-5.5, and Gemini Pro.
Anthropic's Claude Science is a workbench that gives scientists one environment to do computational research, saving them from the need to bounce between databases, pipelines, and tools.
Engineers on the new team will embed within companies to deploy purpose-built agents, focusing on fast deployments and customer self-sufficiency.
Artificial intelligence is transforming what is possible in agriculture, but industry leaders should be wary of investing in AI without first laying the groundwork.  The use cases are promising, especially for an industry navigating volatile fertilizer costs, unpredictable weather, and margins that leave little room for error. Research shows AI-enabled predictive models can improve crop…
Wix-owned vibe coding platform Base44 has started rolling out its own AI model â with hopes that it will eventually outperform frontier models.
Google is expanding Geminiâs personalized AI image generation to eligible free users in the U.S., allowing the chatbot to create images based on your interests and data from connected Google apps.
The startup, Proception, is taking a unique approach to collecting training data to tackle one of the hardest problems in robotics: hands.
Paul Meade, the Apple vice president in charge of the Vision Pro headset, is reportedly leaving the company to join OpenAIâs hardware team.
New models are launching in Asia that promise Mythos-like capabilities without fear of an export ban. U.S. AI labs may never recover this enormous market.
Over 100 companies and government agencies are reportedly authorized to use Mythos 5, including their non-American employees.
Nvidia has dominated the AI chip market for years, but the era of total dependence might be ending.   OpenAI just shared its plans to spice things up with Jalapeño, its custom inference chip built with Broadcom, joining Google, Apple, and SpaceX in a growing list of companies building their way out of single-supplier risk. The goal is less of a […]
Less than 24 hours after news broke that OpenAI would stagger its next model release at the request of the Trump administration, that model, GPT-5.6, is here. On Friday, the company unveiled the limited preview of its new GPT 5.6 model suite: Sol, the flagship; Terra, a medium-tier model for "high-volume work"; and Luna, a […]
Agent-testing startup Patronus AI, founded by former Meta AI researchers, is experiencing nearly insatiable demand, its investor says.
General Intuition has raised $320 million to scale AI trained on millions of hours of gameplay, betting action data can help AI develop something closer to human intuition.
Un-0 is an image-generation system tool that shows for the first time how the company's technology can replicate conventional AI systems.
Adobe said that it will integrate Topaz Labs' tools across its apps.
Amazonâs latest India investment comes as global tech companies race to expand AI infrastructure in the country.
IBM has built a new prototype chip with around 100 billion transistors on an area the size of a fingernail, which is twice the density of the companyâs previous state-of-the-art technology announced in 2021. The design could pave the way for faster and more energy efficient computers for years to come. For more than half…
OpenAI has just revealed a new "intelligence processor" chip for AI servers made in partnership with Broadcom. The chip, called Jalapeño, is designed to power current and future large language models, according to an announcement on Wednesday. Jalapeño is an ASIC (Application-Specific Integrated Circuit), meaning it's designed for a specific purpose: AI inference. With AI […]
AI is booming. New use cases are emerging each day. To capitalize on the technologyâs potential, enterprises require data at scale. In many cases, though, the relevant information is blocked or unstructured, which limits its use by AI models.  To understand this challenge, consider the foundation of the web itself. The web was not designed…
Last week, Midjourney, an AI startup best known for its image generator, made an unusual pivot: medical imaging. The company announced a futuristic ultrasound scanner that would dunk users into a vat of water and, hopefully, produce "something as powerful as MRI" yet "as casual as a trip to the spa." Midjourney says the goal […]
OpenAI is using AI to help the open source community better protect itself.
What does an AI company do after one of those not-acqui-hire deals? Groq raised money, is leaning into its neocloud business, and is hiring new execs.
Reflection AI will pay $150 million a month beginning July 1, 2026 through 2029 for immediate access to Nvidia's latest GB300 AI chips and supporting hardware across SpaceX's Colossus 2 data center near Memphis, Tennessee.
Just as last week was ending, the US government forced Anthropic to pull its two newest models, Fable 5 and Mythos 5, citing national security concerns after Amazon researchers allegedly found a way to bypass Fable 5’s guardrails.  Cybersecurity researchers have since signed an open letter calling the move dangerous, and Anthropic itself noted the same jailbreaks exist in other models. So is […]
Just as last week was ending, the US government forced Anthropic to pull its two newest models, Fable 5 and Mythos 5, citing national security concerns after Amazon researchers allegedly found a way to bypass Fable 5’s guardrails.  Cybersecurity researchers have since signed an open letter calling the move dangerous, and Anthropic itself noted the same jailbreaks exist in other models. So is […]
Startup Baseten is reportedly close to finalizing a $1.5 billion round at a $13 billion as the âinference gold rush" marches on.
AWS is in talks to sell its chips to other data centers. CEO Andy Jassy has said this represents a $50 billion opportunity for the company.
The startup trains embodied AI and world models using Medalâs dataset of 2 billion videos per year from 10 million monthly active users.
Adobe is introducing some new capabilities for its Firefly AI assistant, alongside a "reimagined" AI studio that lets you edit and generate new designs from a single interface. The new Firefly experience launching today in private beta is designed to give you "persistent context, reusable assets, and organized workflows" across your projects, according to Adobe, […]
Adobe's plan to stick AI assistants into all of its Creative Cloud suite is now fully underway, with new chatbots now rolling out to its biggest editing and design apps. As part of a public beta launching today, Photoshop, Premiere, Illustrator, InDesign, and Frame.io now each have a bespoke AI Assistant that can be used […]
Forget stickers, GIFs, and emoji reactions. Pixi is betting that the next evolution of messaging is interactive augmented reality (AR).
Midjourney CEO David Holz just showed off the company's first hardware product and plans to build a San Francisco spa, which he admitted is a bit different from the "cat pictures" produced by its AI image generator. Dubbed The Midjourney Scanner, it's an ultrasound-based full-body scanner that uses a ring of sensors to capture vertical […]
Amazon CEO Andy Jassy may have been the source of security concerns that led Anthropic to cut off worldwide access to two models on Friday.
This article is part of our exclusive IEEE Journal Watch series in partnership with IEEE Xplore. As robots advance in terms of dexterity and other physical capabilities, it becomes more likely that humans may find themselves working alongside them. If that happens, how will robotsâ emotional capabilities need to advance for them to successfully work with people?In a recent study, researchers trained collaborative robots to read human emotions by not only accounting for facial expressions, but also contextual factors in the interactions as well. Through experiments with 40 volunteers, the researchers then evaluated how a robotâs ability to read human emotions and adjust its behavior in turn impacted a humanâs perception of the robot and its capabilities as the two collaborated on tasks. The resultsâwhich show that the emotional capabilities of robots only go so far with humansâwere published 18 May in IEEE Robotics and Automation Letters.Seung Chan Hong led the study as part of his undergraduate thesis while studying at Monash University, in Melbourne, Australia. He notes that, while there has been a lot of hype in the advancing physical abilities of robots, this is only one piece of the puzzle. âWe need to also innovate when it comes to them actually interacting with humans, not just their physical capabilities,â he says.This prompted him to dig deeper into the emotional aspects of human-robot interactions. First, Hong and his co-authors decided to train a robot to read human emotions using a vision language model (VLM), which is similar to large language models (LLMs) such as ChatGPT, but which can also take visual inputs.Training VLMs for Human Emotion RecognitionTo evaluate their VLM, which used Gemini 2.5, the researchers had volunteers watch videos of robots handing over objects to humansâwith varying degrees of successâand describe the emotions the humans were expressing. Importantly, the volunteers labeling these videos were able to take into account more context in these interactions, rather than reporting solely on the facial expressions of the humans in the video. For example, a person pausing to think with a furrowed brow may simply be concentrating on their task at hand and not necessarily be angry. Contextual factors such as drumming their fingers, pursing their lips, or other behaviors can point to the real cause of a personâs furrowed brow.The researchers then compared their VLM to a conventional AI system that relies on standard facial analysis and object tracking that is used in human-robot interactions. They found that the VLM outperformed the traditional approach. On a scale from 0 (no similarity in meaning to the emotion identified by the human volunteers) to 1 (a perfect match in meaning), the conventional AI system achieved a score of 0.77. In comparison, the VLM achieved a score of 0.86.Hong says, âI think [the VLM] was able to align with what human observers were seeing a lot better, because it wasnât just looking at the personâs face for a brief amount of time, but seeing the whole sceneâwhere the person was and what they were doing, and how they were interacting with the robot.âIn a second experiment, the research team asked 40 volunteers to interact with a robot using their VLMâbut purposefully programmed the robot to make an error. The robot then had to offer either an emotionally adaptive apology that accounted for the humanâs perceived response to the mistake or a pre-scripted spoken apology.Participants overwhelmingly preferred the emotionally adaptive response, with 31 out of 40 people favoring this approach over a boilerplate apology.However, their survey responses underscored how this emotional adaptivity was far less important than the robotâs functionality. After collaborating with a robot that failed in its task, many participants ranked their trust in the robot as lower, regardless of how it apologized for its mistake. âA personalized apology acts as a social lubricant, but it cannot repair the trust lost by the robot failing its physical task,â Hong says.Interestingly, the VLM classified the emotions of its human partners similarly to human volunteers who observed an interaction from a third-party perspective. But when the VLMâs assessments were measured against humansâ self-reported emotions during the second experimentâthe most accurate descriptions of their true emotionsâits ability to accurately predict emotions dropped significantly.âWhile the VLM is a good observer of outward social cues, it isnât a mind reader,â Hong says. âIt matched third-person human observers well, but it didnât always align with the usersâ internal, self-reported feelings.âTogether, these results show that robots are not perfect at reading human emotions. So while people might appreciate their efforts, they still ultimately will want competent co-workers.This story was updated on 15 June 2026 to correct where the research was conducted and clarify that the researchers evaluated the performance of a pre-trained model.
The most popular camera in the world just got its first set of serious AI photo editing features, and I don't think any of us are ready. As far as AI photo editing goes, the new features in iOS 27 are pretty tame compared to what you can do on, say, Google's Pixel phones. But […]