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

Researchers at Hacktron utilized Anthropic's Claude Opus 5 to exploit two vulnerabilities, gaining control of OpenAI staff accounts for ChatGPT and Codex. The vulnerabilities were found in the software running OpenAI's public help forum and its login system. By chaining these flaws, the researchers were able to access an internal OpenAI code repository. This exploit was conducted as part of a security research effort. The researchers successfully took over the accounts of several OpenAI employees. The vulnerabilities were exploited in a chained manner to achieve this level of access.