Claude helped hack OpenAI employees' ChatGPT accounts

A team of researchers used the Claude model, specifically the Opus 5 version, to exploit a vulnerability and gain access to OpenAI employees’ ChatGPT accounts. The breach exposed internal data and highlighted weaknesses in the company’s security infrastructure. Experts have called for stronger safeguards against misuse of AI models.

Summary written by Kernelia from the original article by Hipertextual. The story and its rights belong to its author.