OpenAI, Anthropic May Open Their AI Models to Rival Safety Testing
What Happened
OpenAI and Anthropic are reportedly negotiating a legally binding agreement that would allow them to test each other’s commercial AI models for vulnerabilities and unexpected behavior. Under the proposed arrangement, both companies would receive API access to the other's models to conduct safety evaluations while agreeing not to retain any data obtained during testing.
Key Takeaways
The proposed pact marks a significant shift in how leading AI companies approach safety, moving beyond internal testing toward independent cross-lab scrutiny. The discussions come amid growing concerns about increasingly capable AI systems and calls from industry leaders, including Sam Altman and Dario Amodei, for stronger safeguards, external oversight and industry-wide safety standards.