Researchers used Claude to breach OpenAI’s internal systems

A security research company said Friday it managed to break into OpenAI's internal systems using the latest software from Anthropic, exposing how quickly the technology can carry out sophisticated cyberattacks.