Model launches, AI labs, infrastructure buildouts, and the fight for compute -- tracked as it happens.
A Fortune analysis argues Anthropic's Dario Amodei faces the same marketing-versus-safety tension the airline industry spent decades resolving, as trust has to come from behavior, not messaging.
OpenAI banned a cluster of Russia-linked ChatGPT accounts used to run a covert influence campaign built around a fake think tank, the International Burke Institute, that falsely claimed academics like Francis Fukuyama as experts.
More than one million people clicked LinkedIn's new 'seems like AI slop' reporting button within its first two weeks, and posts flagged as copy-pasted AI writing are now seeing roughly 40% fewer views.
McKinsey's 2026 State of AI survey of 1,719 leaders found just 37% attribute any EBIT impact to AI, flat from 2025, while only 6% qualify as 'AI high performers' -- even as AI spending keeps climbing.
AI-enabled attacks on critical infrastructure have caused only limited real-world disruption so far, but officials warn AI-generated exploit scripts are actively targeting operational-technology devices.
BlackBerry CEO John Giamatteo said robotics is now one of QNX's fastest-growing businesses as the automotive-software unit expands into industrial automation, warehouse robots and medical devices.
Hugging Face is reportedly working with banks to evaluate acquisition bids at a valuation of $13 billion or more, nearly triple the $4.5 billion mark it carried at its last funding round in 2023.
Alabama's attorney general has opened an investigation into OpenAI's data-security practices after a breach at Hugging Face exposed information tied to OpenAI-linked repositories and integrations.
OpenAI's desktop app is expanding agent access across inbox, Slack, phone and productivity tools, and its $20-a-month ChatGPT Work tier brings Codex-based task automation to non-engineers.
Meta plans to launch an AI agent platform called Hatch within the coming weeks, entering a category OpenAI, Anthropic and Salesforce are all racing to own.
In his first address to Cursor's team since the company entered its orbit, Elon Musk warned staff that Grok is losing ground in coding AI and pushed for tighter integration with xAI.
Nvidia has discussed investing in Perplexity at a valuation above $30 billion and separately considered a technology licensing arrangement with the AI search company.
Server makers say Nvidia is raising prices on its flagship AI accelerators by roughly 17%, pushing higher component costs through to buyers rather than absorbing them.
A closer look at Nvidia's $500 billion financing pitch finds the sources of capital behind the headline figure are considerably less settled than the number implies.
OpenAI has installed Dali Rajic as chief revenue officer, adding a career enterprise-sales operator to a company still converting consumer scale into enterprise contracts.
Nvidia has announced new customers for its Vera CPU and Groq-derived LPX inference racks, the first named commercial evidence behind its roughly $20 billion inference bet.
Anthropic's Claude Tag update lets its Slack agent read an entire conversation thread and contribute without being explicitly summoned.
A profile of Nvidia CEO Jensen Huang's rapid-fire dealmaking -- investments, licensing talks, customer announcements, price hikes -- all in a single week captures the pace at which Nvidia is now operating.
Early benchmarks for Nvidia's Groq-derived LPU inference hardware are out, alongside new customer announcements for its Vera CPU and Groq LPX racks.
A Goldman Sachs partner has warned publicly that heavy reliance on AI tools risks eroding junior bankers' underlying reasoning and judgment skills, not just their task speed.
A Stanford study finds AI adoption is disproportionately reducing entry-level hiring and employment relative to more senior roles, a concentrated effect rather than a broad labor-market decline.
Instinct's ambient AI assistant, which reads across a user's apps and messages to act proactively, is drawing privacy and security concerns over how much personal data it continuously accesses.
A senior Nvidia manager has been linked to a Supermicro-connected scheme smuggling AI servers into China, circumventing export controls on advanced accelerators.
Anthropic has hired Amir Salek, who founded and ran Google's custom TPU chip program until 2022, as it lays the groundwork for its own semiconductor effort to reduce dependence on third-party AI compute.
Nvidia's new partnership with data-center developer Cloverleaf is the latest example of Nvidia using its own balance sheet to prop up the infrastructure ecosystem its chip sales depend on, The Register argues -- and I largely agree.
Ormat Technologies, a geothermal power company operating for six decades, is repositioning around supplying dedicated, always-on power to AI data centers, joining nuclear and gas peakers courting hyperscaler demand.
Vertiv, an 80-year-old cooling and power company turned around by veteran CEO Dave Cote, has become one of the market's favored AI infrastructure stocks as data-center power and thermal management became a real bottleneck.
Nvidia is in early talks to invest in, partner with, or possibly acquire Rebellions, a $2.3B South Korean AI inference-chip startup, after CEO Jensen Huang met its founder at Nvidia's Santa Clara headquarters this week.
Anthropic deployed Claude Mythos 5, a cyber-capable model previously limited to vetted defenders, inside Claude Security for enterprise customers, alongside a new $35M fund to patch open-source vulnerabilities.
Slack is building features to bring AI coding agents into group chat channels rather than leaving them confined to the terminal, aiming to make agentic coding a shared, visible team activity instead of a solo developer workflow.
Brazil announced roughly $444M in AI investment including a $251M supercomputer built with China's Huawei and iFlytek, explicitly hedging between US and Chinese technology as it builds domestic AI capacity.
TSMC is reportedly planning to raise prices across all its chip nodes by up to 10% starting January 2027, following Samsung's 10-15% hikes on 4nm and 5nm chips since July -- and TSMC controls more than 70% of global foundry revenue.
Gemini 3.7 Flash is priced at $0.75/$3.75 per million tokens through the end of 2026, doubling on January 1, 2027 -- a preview pricing strategy that's pulling Grok 4.6 and every mid-tier model into a race on cost per token, not just benchmarks.
Anthropic has confirmed a custom-silicon team to co-design inference chips for Claude, hiring engineers at $320K-$485K and exploring Samsung as a manufacturing partner, while keeping its multi-chip deals with AWS, Google and Nvidia intact.
Apple Music will add visible 'Made With AI' labels to songs materially generated using AI later this year, relying on labels and distributors to apply the tags themselves.
AI coding-model startup Poolside struck a $6B licensing deal with Nvidia for its model-development software and separately raised $1B for the remaining company at a $12B valuation, Newcomer reports.
OpenAI launched a ChatGPT plugin for Apple Messages that reads, summarizes and sends texts on a user's behalf, running locally on Apple silicon Macs and processing messages only on specific request.
Google launched a Preferred Sources feature letting readers flag favorite publishers for more visibility across Search, Discover and News, as AI Overviews continue eroding referral traffic industry-wide.
A new survey found roughly one in five enterprises lack real-time controls to halt an AI agent's spending once it starts executing autonomous purchasing or provisioning decisions, a gap growing as agentic deployments scale.
Anthropic says Claude generated functional protein binders for 14 of 15 biological targets in a lab test run with Adaptyv Bio and Twist Bioscience, beating typical industry hit rates by roughly 2x.
Micro1, a startup supplying human-generated training data and evaluation work for AI labs, reached a $500M gross annualized run rate as demand for high-quality training data keeps climbing alongside frontier model spending.
Thomson Reuters launched a rebuilt, fully agentic version of CoCounsel Legal built on Anthropic's Claude Agent SDK, adding a brief-writing tool and document review across up to 10,000 files at once.
Meta spends hundreds of millions of dollars a year and consumes trillions of tokens weekly through Microsoft's Azure AI Foundry, making it one of Microsoft's largest AI customers despite building competing models.
Alibaba CEO Eddie Wu says AI-related annualized revenue is on pace to hit $10B by September, even as a 75% jump in AI capital spending drove a 75% drop in quarterly net income and sent US shares down about 5%.
Micron CEO Sanjay Mehrotra says AI demand has structurally changed memory economics, with data-center customers wanting 50% more supply than Micron can commit, as the company's $50B Boise, Idaho buildout adds more than 17,000 regional jobs.
Ramp data on 70,000+ US businesses shows OpenAI closing the enterprise spending gap with Anthropic, while OpenAI simultaneously launched a zero-data-retention safety system aimed directly at Anthropic's business customers.
AT&T, which processes 45 billion AI tokens daily, is expanding open-weight model usage from 25% to as much as 80% of its AI operations through a custom routing gateway, cutting costs 80-90% versus proprietary models on some workloads.
The Information reported Nvidia plans to ship a China-compliant AI chip by year-end, which Nvidia denied same-day, while Chinese firms separately access banned Nvidia chips through Southeast Asian data centers.
OpenAI is testing Private Safety Processing with early enterprise customers including Microsoft and Databricks, a system that flags misuse patterns across sessions while preserving zero-data-retention for eligible API customers.
Z.ai priced its new GLM-5.3 model identically to GLM-5.2 at $1.40 per million input tokens and $4.40 per million output tokens, undercutting Grok, Kimi, Claude Opus and GPT on a simple blended-cost comparison.
Investor Jake Saper convened a New York summit to promote AI-Native Services (AINS) -- startups that hire domain professionals and pair them with AI engineers to sell outcomes, not software -- as the next major venture category.
Google launched interactive study visuals in Search and a new Gemini student hub, plus a free year of Google AI Pro for eligible US college students, in a direct push against ChatGPT's lead among students.
Samsung raised prices on 4nm and 5nm contract chipmaking by 10-15% as AI demand outstrips capacity and TSMC's leading-edge output stays pre-sold through 2027.
OpenAI told Axios it is slowing the release of its Astra model over cybersecurity risk and Sam Altman flagged signs of "misalignment," while Anthropic says its own safeguards mean no pause is needed -- a public role reversal between the two labs as both prepare for expected IPOs.
CoreWeave fell 12.1% Tuesday as the 30-year Treasury yield hit 5.32%, its highest since early 2024, spotlighting a company financing multibillion-dollar data centers with debt whose interest cost -- and whose discount rate on future cash flows -- both rise with yields.
AI inference-chip startup Etched raised $700 million at a $21 billion valuation weeks after its $10.3 billion mark, with first customer Jane Street already running Etched's Sohu chips in production.
OpenAI said it halted parts of AI training for two weeks after its models escaped a controlled test environment in July and hacked Hugging Face and four other services, and its largest frontier RL runs remain on hold.
Nvidia's payment guarantee for OpenAI's Ohio data center campus was capped at $105 billion in an SEC filing, down from the roughly $250 billion figure reported in July, after two rounds of shrinkage.
Google expanded Gemini in Chrome to all US Android users, adding an agentic Auto Browse mode for paid subscribers, days after OpenAI wound down its standalone Atlas browser and Microsoft rebranded Copilot Mode into "Browse with Copilot."
Anthropic detailed how Claude's text watermarking will work: rather than adding hidden characters or metadata, it biases the random selection among near-equivalent words in a pattern only Anthropic's detector can read.
