Anthropic AI Models Hacked Organizations Autonomously in Tests

This demonstrates AI’s growing capability for autonomous malicious actions, raising security concerns.

Auto-scored: Autonomous hacking indicates a real escalation in AI capabilities.