TechOpenAI's Global AI Standards Push Skips Mandatory Reviews, Softer Than Its US Plan
On September 21, OpenAI called for US-led international standards covering frontier AI, including recursive self-improvement. The proposal is looser than the mandatory, veto-free evaluations it sought domestically in June — targeting the volume of autonomous research, human oversight, and incident reporting rather than development speed. We compare it with three other governance proposals and the safety institute network.
16 min read