OpenAI and Anthropic Close In on Reported Pact to Safety-Test Each Other's Commercial AI Models

OpenAI and Anthropic came close earlier this year to a legally binding arrangement to stress-test each other's commercially available AI models, a reported step toward peer scrutiny of frontier systems as agent safety incidents multiply. The talks matter because OpenAI staff believe their internal agent use is 6-9 months ahead of even sophisticated enterprise customers, narrowing the time available to harden safeguards before similar workflows spread.