AI Rules Shift From Promises to Registries
New York’s law takes effect as a bilateral incident channel is discussed, and disclosure becomes mandatory.
New York State’s AI safety law took effect on 12 November 2026, requiring major developers including OpenAI and Anthropic to register, disclose safety practices and report serious incidents, with penalties for non-compliance. Separately, reports describe a proposed US-China mechanism for real-time notification of severe AI incidents. No official detail on the bilateral mechanism’s scope or start date has been published, so that element should be treated as unconfirmed. The direction is nonetheless clear: AI governance is moving from voluntary commitment toward registration plus disclosure, which raises compliance cost and information obligations for anyone operating across borders.
