OpenAI is calling for shared international standards for the most capable AI systems. In a September 21, 2026 policy post, the company proposed using existing safety institutes to develop technical rules, with the United States leading the effort.
One focus is research in which AI helps develop new AI systems. OpenAI warns that increasing automation could outpace human oversight. It says fully autonomous recursive self-improvement is not happening today and should not be pursued until it can be done safely.
The proposal covers comparable measures of research progress, triggers for human review and common reporting rules for misalignment incidents. Technical coordination would include open and closed models. National governments would decide whether to incorporate the standards into their own laws.
This does not establish an international regulator. OpenAI explicitly distinguishes the proposed standards from model licensing and mandatory approval procedures. There is no binding timetable. The next questions are which institutions take part and what verifiable obligations follow. The post states a safety-policy position; it does not document another loss-of-control incident.