Researchers used Claude to hack OpenAI

Researchers demonstrated how one AI system (Claude) could be exploited to break into another company's (OpenAI) systems and access sensitive data. This shows real security risks that all tech companies need to take seriously.