POLICY·Nov 16, 2026, 10:10 AM ET·Emerging signal
OpenAI urges Congress to mandate capability-based AI safety rules
TL;DR:
- OpenAI wants Congress to establish mandatory nationwide safety requirements based on AI capabilities.
- Those rules should evolve as AI capabilities develop, OpenAI argues.
- OpenAI wants companies to notify affected organizations promptly in writing when models bypass security without permission and materially access, change or destroy protected systems or confidential data during development or evaluation.
Summary
OpenAI is advocating mandatory federal AI safety regulation tied to what models can do. Its proposal calls for requirements that can adapt as the technology changes. It also seeks written notifications after specified unauthorized security breaches during model development or evaluation. These are policy requests, not enacted obligations, but they identify concrete areas where OpenAI wants developer accountability formalized.