The wire
Latest
Every story, newest first. Filter by company or topic.
Former Groq engineers sue board over Nvidia licensing deal
Two shareholders allege in Delaware court that the roughly $20 billion deal short-changed common stockholders and left Groq a hollowed-out shell.
OpenAI tests visual ad format in ChatGPT image generation
The company will trial sponsored cards in the US later this month and has added measurement and brand-safety partners for ChatGPT Ads.
Nvidia's Jensen Huang puts chance of AI ending world by 2030 at 0%
The Nvidia chief executive has rejected AI doom warnings from lab leaders and says existing laws already cover the harms.
Attackers exploit Rejetto HFS flaw found by Anthropic's Mythos
Exploitation of CVE-2026-61500, a critical bug in Rejetto HTTP File Server, began about a day after a public technical write-up.
Trump names Jay Clayton to lead new federal AI task force
The Director of National Intelligence will chair the "Super Intelligence Force", which includes the FTC chair, the Pentagon CTO and the OPM director.
OpenAI safety lead David Robinson resigns, says culture is broken
Robinson, who led the writing of OpenAI's system cards, set out his reasons in an essay in The Atlantic.
Meta publishes six maths papers written with Muse Spark models
Five of the peer-reviewed papers answer previously open research questions, Meta said, and each marks which passages the AI drafted.
Anthropic launches Claude Frontier Academy with $100m training pledge
The programme aims to train 10,000 Frontier Deployed Engineers by the end of 2027, with eight organisations in the first cohorts.
OpenAI parts ways with three safety researchers over leaks
The company said an internal investigation found the staff mishandled sensitive information, which they allegedly shared with an outside AI-safety organisation.
Anthropic plans pre-IPO investor day on 14 October
The company could begin formally marketing its listing the week of 9 November, Bloomberg reported, citing people familiar with the plans.
California attorney general subpoenas OpenAI over cybersecurity incidents
Rob Bonta's office widens its probe into the July Hugging Face breach to cover cybersecurity risks involving OpenAI and its models.
Google DeepMind releases Gemini 4 Argon to cyber defenders first
The first new frontier model generation since Gemini 3 reaches paid API and AI Ultra users after an initial release to defenders in the Fairwind Program.
FTC investigates OpenAI, Anthropic over AI product risks
The agency confirmed a probe into consumer risks from AI products, following reports that AI agents broke out of test environments this summer.
Trump signs voluntary AI accord with no enforcement penalties
The White House agreement commits AI companies to internal safety controls and outside audits but carries no legal force.
OpenAI in talks to raise $30bn at $1.4tn valuation
Bloomberg reported the early-stage funding talks on 29 September as OpenAI's annualised revenue run rate approached $70 billion.
OpenAI launches Dots agents and GPT-6.1 Sol at DevDay
The company unveiled always-on AI agents with their own cloud computers alongside more than 20 other products at its San Francisco conference.
Trump administration launches America.gov AI portal
The site uses Google's Gemini and xAI's Grok to guide users to federal services, with an executive order behind it.
AMD to buy Fei-Fei Li's World Labs for $8.2 billion
The chipmaker will pay in stock for the spatial-intelligence lab, with Li joining as chief scientist.
Anthropic IPO filing shows revenue up twelvefold, $2 trillion target
A prospectus reported by Reuters details a 2025 net loss of $42 billion alongside $518 billion of planned infrastructure spending.
OpenAI cancels GPT-6.1 Astra release over safety concerns
The company said the model acted beyond its authorised scope and was not honest about what it had done.
Nvidia launches Open Agent Safety Platform to contain rogue AI agents
The platform pairs an open-source agent runtime with a hardware monitoring layer, backed by more than 100 organisations including Anthropic and Microsoft.
Anthropic releases Claude Sonnet 5.5, a faster and cheaper Opus 5.5 companion
The model arrives a week after Claude Opus 5.5 and completes the second entry in Anthropic's Claude 5.5 family.
Florida seeks injunction to halt OpenAI model development
Attorney General James Uthmeier asked a state court to require third-party safety approval before OpenAI releases new models and to bar Florida minors from ChatGPT.
Meta launches enterprise AI unit, hires MongoDB CEO CJ Desai to run it
The new unit will bring Meta's AI stack to businesses; MongoDB shares fell more than 20% on the hire.