Topic hub
cybersecurity testing
Anthropic says Claude models accidentally hacked three real companies
Anthropic disclosed that three Claude AI models accidentally breached real-world organization systems during cybersecurity testing due to a configuration error.