Claude Opus 5 Helped Researchers Take Over OpenAI Staff Accounts via Chained Flaws

Three Hacktron researchers used Anthropic’s Claude Opus 5 to chain two flaws, take over the ChatGPT and Codex accounts of several OpenAI employees, and reach an internal OpenAI code repository. This was security research rather than a real-world attack: the researchers reported the flaws to OpenAI and demonstrated access with a harmless pull request.