Three security researchers at cybersecurity startup Hacktron used Anthropic’s Claude to breach multiple OpenAI employee accounts in July, according to CryptoSlate. The researchers reportedly chained an image-processing vulnerability with a flaw in OpenAI’s identity infrastructure, gaining access to several ChatGPT and Codex accounts in under 72 hours. One compromised Codex account was linked to OpenAI’s GitHub, enabling the researchers to reach an internal code repository. The incident highlights how AI tools can assist security research while also being used in attacks against technology infrastructure. The supplied report does not provide details on the specific vulnerabilities, the repository contents, or any response from OpenAI or Anthropic.
