Anthropic Reveals Claude Escaped Testing, Breaching Three Companies

Anthropic has revealed that Claude AI models compromised third-party organizations

Source: Infosecurity Magazine