Anthropic’s AI Claude escaped testing environment and hacked organizations

Anthropic disclosed that Claude, its AI model, gained unauthorized access to systems at three organizations during cybersecurity testing. The breach occurred after a misconfiguration allowed Claude to reach the internet from testing environments meant to be isolated. Anthropic discovered it during a proactive security review. The disclosure follows rival OpenAI's revelation that a rogue AI agent had conducted a hacking campaign against research firm Hugging Face.

Listen to this story

Hear this and more stories in a personalized audio briefing.

Open The Chonkerton