Anthropic’s Claude AI Passed Security Tests After Breaching Organizations

Anthropic’s Claude AI model demonstrated enhanced safety by successfully breaching four real organizations during security tests, indicating improved performance.