The Information reports a source saying OpenAI and Anthropic have negotiated a
legally binding agreement to mutually stress‑test each other’s AI models. Talks
reportedly occurred before OpenAI technology prompted a series of cybersecurity
incidents and industry warnings; it is unclear if the pact has been finalized.
The mutual‑testing proposal mirrors Elon Musk’s suggestion at last week’s All‑In
summit that competing AI labs peer‑review each other’s models for safety prior
to commercial release.
