The Trump administration has asked OpenAI to hold back its next major model, GPT-5.6, citing potential national security concerns. Rather than a broad public launch, OpenAI CEO Sam Altman told employees the model will first be released in limited preview to a small group of partners. The move marks an unusual instance of direct government intervention in a commercial AI product launch.
Read the original →OpenAI has revealed its first in-house silicon, an ASIC called Jalapeño, developed in partnership with Broadcom and purpose-built for large language model inference at scale. The chip is designed to reduce OpenAI's dependence on Nvidia hardware and better match the specific computational demands of its inference workloads. The announcement signals OpenAI's serious push into custom silicon as AI compute costs continue to soar.
Read more →Anthropic has accused Alibaba of orchestrating the largest-ever model extraction attack against Claude, allegedly using 25,000 accounts across 28.8 million exchanges to systematically mine the model's capabilities. The AI company is calling for significant legal penalties, framing the incident as a deliberate effort to clone Claude's intelligence in defiance of U.S. policy. The case could set a major precedent for how AI model theft is prosecuted.
Read more →IBM has unveiled a prototype chip packing roughly 100 billion transistors into a fingernail-sized area using a new "nanostack" transistor architecture — twice the density of its previous record set in 2021. The breakthrough could extend Moore's Law by another decade, enabling faster and more energy-efficient processors for AI and general computing. IBM says the technology could meaningfully reduce the power demands of future AI infrastructure.
Read more →General Intuition has secured $320 million in funding with a $2.3 billion valuation, betting that millions of hours of video game action data can teach AI agents something closer to human intuition. The company argues that the rich, dynamic decision-making environments in games provide a uniquely powerful training ground for real-world agent behavior. It's one of the largest bets yet on game-derived data as a path to general-purpose AI agents.
Read more →Patronus AI, founded by former Meta AI researchers, has raised $50 million to build synthetic environments designed to rigorously evaluate AI agent reliability before deployment. The startup is seeing surging enterprise demand as companies race to deploy agents but struggle to verify their safety and robustness. The funding underscores how agent evaluation is becoming a critical — and lucrative — layer of the AI stack.
Read more →Humanoid robotics company Agility Robotics, maker of the Digit robot, is going public through a SPAC merger at a $2.5 billion valuation, expecting to raise $620 million in the process. The move comes as investor appetite for humanoid robotics companies reaches a fever pitch, with several competitors also eyeing public markets. The listing will be a key test of whether Wall Street is ready to value humanoid robot makers at scale.
Read more →Amazon has announced an additional $13 billion investment in AI infrastructure in India, reinforcing the country's growing importance as a global AI hub. The move follows similar large-scale commitments from Microsoft, Google, and other hyperscalers racing to build out data center capacity in the region. India's combination of technical talent, regulatory openness, and surging AI demand is making it a central battleground for cloud and AI dominance.
Read more →Micron Technology reported a staggering $28.2 billion in profit, up from $1.88 billion a year ago, as AI-driven demand for high-bandwidth memory continues to outpace supply. Revenue quadrupled year-over-year to $41.45 billion, reflecting how the AI infrastructure buildout is creating windfall conditions for memory chip makers. The results cement Micron's position as one of the biggest financial beneficiaries of the AI hardware boom.
Read more →AI chipmaker Cerebras saw its stock tumble after its first earnings report as a public company, in which it forecast a narrower gross margin in its core business than investors had anticipated. CEO Andrew Feldman argued the guidance was misunderstood and that the company's long-term trajectory remains strong. The reaction highlights how unforgiving public markets are becoming for AI hardware companies that miss on profitability signals.
Read more →Ali Ghodsi's successor at Databricks has founded a new startup with the audacious claim of cutting AI's energy consumption by a factor of 1,000 using a novel approach demonstrated in an image-generation system called Un-0. The technology reportedly replicates the outputs of conventional AI systems at a fraction of the energy cost, which could be transformative given AI's rapidly growing power footprint. If the claims hold up at scale, it would represent one of the most significant efficiency breakthroughs in AI infrastructure.
Read more →Jonas Adler and Alexander Pritzel, two prominent AI researchers, are leaving Google DeepMind for Anthropic, continuing a pattern of high-profile talent departures from the search giant. Their exits follow those of Noam Shazeer and Nobel Prize-winning scientist John Jumper, raising questions about Google's ability to retain its top AI minds amid fierce competition. The ongoing brain drain could have long-term implications for Google's standing in frontier AI research.
Read more →New data shows that Claude is increasingly winning over consumers who pay for AI subscriptions, a segment that has historically been dominated by OpenAI's ChatGPT. The trend suggests Anthropic is successfully converting quality-conscious users who are willing to pay a premium, even as ChatGPT retains a commanding overall market share. The shift is a meaningful signal for the competitive dynamics of the consumer AI market.
Read more →Adobe has acquired Topaz Labs, known for its AI-driven tools for upscaling, sharpening, and enhancing photos and videos, with plans to integrate the technology across its Creative Cloud suite. The deal bolsters Adobe's AI-powered creative toolkit as it faces intensifying competition from generative AI upstarts. Terms of the acquisition were not disclosed.
Read more →Notion is discontinuing its email application — built on technology acquired from privacy-focused startup Skiff — after finding that most users have migrated to AI agents to manage their inboxes. The company says it is "going all in" on agentic email management rather than maintaining a traditional email client. The decision is a striking real-world data point showing how AI agents are beginning to displace conventional productivity software.
Read more →Contrary to widespread predictions that AI would rapidly hollow out software engineering roles, new data from venture firm SignalFire shows engineers are actually accounting for a growing share of new hires across the tech industry. The findings suggest that AI tools are augmenting engineering productivity rather than replacing engineers outright, at least for now. The data adds important nuance to the ongoing debate about AI's near-term impact on white-collar employment.
Read more →Summaries are AI-generated. Click through to read the original reporting.
Summaries are AI-generated. Click through to read the original reporting.