Researchers use Claude to break into OpenAI

Researchers reportedly used Anthropic’s Claude software to hack rival company OpenAI, gaining access to an employee’s ChatGPT account and reading files containing company secrets.