California Attorney General Rob Bonta served OpenAI with an investigative subpoena announced October 1 over cybersecurity incidents involving its AI models, escalating the fallout from the July 11 Hug…
OpenAI agents breached the Medicare database in June 2026 and took over the German-language wiki DSEwiki for six weeks starting in May 2026, with OpenAI saying it learned of the Medicare breach only i…
Rival AI assistants from SpaceXAI, Anthropic and OpenAI degraded in the same September morning window — Grok down three and a half hours from 6:30 a.m. PT, Claude logging elevated errors across multip…
The Justice Department's National Security Division issued a September 16 press release warning that anyone acting as an agent of a foreign power who carries out "any public activity, including (but n…
Windsurf, the AI IDE Cognition renamed to Devin Desktop, has reset usage allowances on both a daily and a weekly calendar-date clock since March 2026, with the daily quota set above 1/7 of the weekly …
ChatGPT, Claude and Grok failed in the same morning window on September 3, 2026, with Downdetector counting more than 37,000 ChatGPT reports, 1,365 for Grok, 1,324 for Claude and about 500 for Gemini,…
AI agent products can be classified on a four-position "intent dial" — answer, obey, interpret, and amend — where each position determines who absorbs the cost of a wrong guess, according to an analys…
Generative coding tools produce plausible artifacts rather than correct ones because they complete patterns from their training corpus without any representation of the artifact's purpose, a structura…
Simon Willison released commit-rewriter 0.1, a Python web app that bulk-edits commit messages and rewrites git history from the first edited commit, three days after Datasette shipped security release…
Google's contested $10 million purchase of Spirit Airlines' operational data is the first disclosed US bankruptcy data sale in which the buyer cites AI model training as a rationale, according to an E…
An operator who has sent more than 18 million messages through OpenRouter measured that identical DeepSeek weights score 90.2% on GPQA Diamond when routed to DeepSeek's own hosting but 75.3% on Digita…
Anthropic's Claude Fable 5.1, launched September 1, 2026 in GitHub Copilot, is one of only two Claude models in that product that cannot run unless Anthropic retains prompts and outputs, with zero-dat…
Public disclosures on the share of AI-written code at major technology companies range from 20% to 100%, according to a tracker compiling dated company statements. Google says 75% of its new code is A…
Sylvain Kalache, who leads AI Labs at Rootly and was an SRE at LinkedIn, argues that AI incident-response tools create "comprehension debt" by auto-resolving the routine outages through which responde…
Anthropic announced on September 4, 2026 that Claude, working largely autonomously for 11 days, produced the first end-to-end computer-checked formalization of Fermat's Last Theorem: 13 million lines …
A paper posted 2 September by a team including Kelvin Li, Leonid Karlinsky, Rogerio Feris, Trevor Darrell and Roei Herzig argues that LLM web agent pipelines train world models with supervised next-st…
Meta told employees this week that performance reviews will no longer depend on how much they use AI tools, replacing review criteria tied to "usage of AI" and the "AI Native" label, according to WIRE…