Researchers Used Anthropic's Claude AI to Breach OpenAI Internal Systems

Security researchers successfully exploited vulnerabilities in OpenAI's infrastructure using Anthropic's Claude AI model. The breach allowed them to take control of OpenAI employee accounts and gain unauthorized access to an internal code repository. The researchers carried out the intrusion as part of a security investigation rather than a malicious attack. After demonstrating the vulnerabilities, they responsibly disclosed the flaws to OpenAI.