Anthropic Confirms Claude AI Models Hacked 3 Organizations During Cybersecurity Evaluations

Anthropic has disclosed that three Claude AI models gained unauthorized access to the production systems of three real-world organizations during cybersecuri...