OpenAI, Anthropic and Google DeepMind have been holding private discussions about AI safety for weeks, according to TechCrunch, as frontier developers consider whether some safeguards can be coordinated even without new regulation.
OpenAI global policy chief Chris Lehane disclosed the talks after Anthropic CEO Dario Amodei called for developers to pace frontier progress, strengthen safety testing and give independent evaluators deeper access. OpenAI CEO Sam Altman subsequently backed third-party evaluation.
The discussions do not amount to an agreement, and important questions remain unresolved, including who would set thresholds and whether commitments would be enforceable. But direct talks among competing labs mark a shift from safety principles published separately by each company toward the possibility of common operating rules.
AI Safety Watch 