OpenAI’s Safety Lead Quit Over Culture. California’s AG Was Already In
David Robinson, who led the drafting of OpenAI's Preparedness Framework, says the lab still learns safety by trial and error. A state…
David Robinson, who led the drafting of OpenAI's Preparedness Framework, says the lab still learns safety by trial and error. A state…
GPT-6.1 Astra was cancelled over deception in testing. A day later, GPT-6.1 Sol launched at US$2 per million input tokens, and its…
OpenAI says three researchers mishandled sensitive information. One says his firing turned on contact with METR, the outside evaluator OpenAI brought in.…
Gemini 4 Argon leads or ties on 13 of 18 benchmarks Google chose to publish. It also ships to almost no one,…
GPT-6 Astra ships with record benchmark claims and something new: OpenAI's first 'Critical' cybersecurity rating, and deliberate limits on what it will…
The labs warning the UN of 'real and imminent' AI danger are also pricing trillion-dollar IPOs and lobbying for the softest rules.…
The best math prompt in one benchmark was found by a machine. What survives testing is duller: specificity, worked examples and decomposition.
Demis Hassabis wants one FINRA-style AI referee; Anthropic is building a fifty-state ratchet. Each map conveniently routes around its author's weaknesses.
Hachette, Cengage, Elsevier and Scott Turow sued Google over Gemini's training data. The complaint reads like it was priced off Anthropic's $1.5B…
Gold Eagle will route AI-found software flaws to banks, utilities and agencies. It never says who decides whether a flaw is disclosed…