OpenAI and Anthropic have entered negotiations for a legally binding agreement to stress-test each other’s commercial AI models. The proposed deal grants each company API access to identify safety vulnerabilities and unexpected behaviors in the rival's systems.
Under the terms, neither firm would retain the other's data. This collaboration follows public calls for increased caution and independent evaluation from OpenAI CEO Sam Altman and Anthropic CEO Dario Amodei.
While the final status of the agreement remains unclear, the talks represent a significant shift in AI security strategy. The competitors are prioritizing systemic risk mitigation despite potential antitrust scrutiny.