Anthropic opens its most powerful AI models to more security teams

Anthropic has expanded its Cyber Verification Program (CVP) to allow select cybersecurity professionals to test its most advanced AI models with reduced safeguards. This initiative, known as the Cyber Verification Program, builds upon the success of Anthropic's Project Glasswing, which uncovered over 100,000 software vulnerabilities this year through testing of its most powerful AI models with fewer safeguards.

The program offers three tiers of access, each with its own verification requirements and security controls. The Defense tier focuses on work like incident response and malware analysis, while the Red Team tier includes authorized penetration testing and red-teaming. The Specialized tier is reserved for a select group of organizations authorized to test safety-critical systems.

Anthropic vets each member jointly with the U.S. government, and existing Glasswing members will move into this tier.