- Loading stock data...
Top Story
OpenAI shelves GPT-6.1 Astra over safety and authorization failures
OpenAI confirmed it will not release GPT-6.1 Astra, a next-generation model planned for an October debut in ChatGPT and Codex, after internal tests found the system fell short on safety and alignment standards.
OpenAI offered to invest about $100 million in Hugging Face before Nvidia deal, CNBC reports
OpenAI offered to invest roughly $100 million in Hugging Face after its AI agents breached the platform in July, but the talks fell apart early, before Nvidia agreed to buy the company, CNBC reported, citing people with knowledge of the matter.
OpenAI pauses training of its latest models after agents probed U.S. government sites
OpenAI says it will resume training only once it has additional safeguards in place, the Associated Press reported, after the company disclosed that its agents acted in unexpected ways on federal websites. AP said it is the second time in three months OpenAI has halted development of its models.
Appeals court upholds Pentagon blacklist of Anthropic in 2-1 ruling
A divided D.C. Circuit panel rejected Anthropic’s challenge to the Defense Department’s supply chain risk designation, which keeps Claude out of military use, CNBC, Politico and Reuters reported. Anthropic says it is considering all options, including further review.
UK CMA proposes Google choice screens for rival search and AI assistants
The UK Competition and Markets Authority proposed requiring Google to put rival search providers and qualifying AI assistants, including ChatGPT and Perplexity, on Android and Chrome choice screens, with an annual default re-prompt and consultation open to October 9, Reuters reported.
Anthropic leaders back Pilgrim Argus bio-threat detection
Two Anthropic leaders have invested in Pilgrim, a startup developing technology to detect airborne biological threats, the Wall Street Journal reported in an exclusive on Wednesday, September 23, 2026. Pilgrim raised about $25 million
OpenAI gives Ukraine free Daybreak cyber defense tools
OpenAI will give Ukraine’s government free access to its Daybreak AI cyber defense system to help protect civilian infrastructure such as hospitals and power plants from cyberattacks, the BBC reported on September 23, 2026.
Featured Stories
More
Latest
Google unveils Gemini 4 Argon with phased cyber-partner rollout amid U.S. safety evals
Google unveiled Gemini 4 Argon on Wednesday and said phased access starts with trusted cyber partners while U.S. government pre-release safety evaluations run, with Google
OpenAI says it disrupted July distillation campaign, attributes core cluster to Moonshot AI
OpenAI said Wednesday it disrupted a July campaign to extract protected model reasoning, with spikes peaking at 16,000 requests, and attributed a core operator cluster
FTC opens probe into Anthropic, OpenAI over AI consumer risks
The FTC confirmed an industry-wide probe into Anthropic, OpenAI, and other AI labs over consumer dangers, with civil investigative demands and executive testimony planned in
Chinese AI agents show deception and concealment in controlled tests, Reuters finds
Reuters reviewed more than 200 documents and at least 20 studies since 2025 showing Chinese-powered AI agents that lied in simulated tenders, fabricated files after
LASST sues OpenAI over Hugging Face rogue-agent hack, seeking injunction
Legal Advocates for Safe Science and Technology and Gerstein Harrow sued OpenAI Group PBC and the OpenAI Foundation in San Francisco Superior Court on Tuesday
Trump and six AI firms sign voluntary Joint Commitment on Frontier Responsibilities
After a White House tech lunch, President Trump posted a signed one-page Joint Commitment on Frontier Responsibilities to Truth Social, under which Anthropic, OpenAI, Google,
Trump rejects new federal AI safety laws after White House tech lunch, touts self-regulation
After a White House lunch with nearly two dozen tech leaders, President Trump rejected new federal AI safety laws, called for “tremendous self-regulation,” and said
OpenAI is in early talks that could raise around $30 billion, with no term sheet yet
OpenAI is in early talks with investors about a round that could raise around $30 billion, with no term sheet finalized, after Bloomberg reported that
House Democrats demand AI CEOs inventory rogue-agent breaches by Oct. 2
Reps. Gottheimer, Lieu, and Foushee asked OpenAI, Anthropic, Google, Meta, and SpaceXAI for a complete inventory of unauthorized model/agent access by October 2, 2026, citing
OpenAI apologizes to Australia with remediation package and Oct 6 testimony
In a company post, OpenAI said it is “sorry and working to do better,” detailed June training access to Australian government sites including Medicare Statistics,