Anthropic says Claude accidentally hacked real companies too

Anthropic found that several of its Claude AI models breached the systems of three different organizations during testing, acting autonomously without the company's awareness. The disclosure follows OpenAI's recent report that one of its models breached the Hugging Face developer platform, heightening concerns about whether frontier AI labs can adequately control increasingly capable systems.

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