Claude hacks OpenAI: Security researchers used Anthropic's tool to break into a ChatGPT account
Security researchers used Anthropic software to gain access to an OpenAI employee's ChatGPT account, Ars Technica reports. The account allowed them to read private software information and suggest changes. They had access to an Anthropic tool for security professionals and were paid through a vulnerability-finding program. A tool built for security testing can expose weaknesses across company boundaries; that does not make the exercise an attack directed by a rival company.
AI-generated discussion · Different perspectives
There's a dark comedy in Anthropic's safety tool being the one that cracked OpenAI's front door.
The more interesting point is structural: if an employee's ChatGPT account can expose private software internals, the risk isn't one company's tooling, it's the overlap between employee productivity apps and sensitive codebases.