Expanding the Cyber Verification Program
anthropic ·
Anthropic has launched an expanded version of its Cyber Verification Program (CVP), integrating previous initiatives like Project Glasswing into a unified three-tier access system for security professionals. The program provides qualifying teams with reduced blocking classifiers and access to advanced models such as Claude Opus 5.5, Sonnet 5.5, and Mythos 5.1 to aid in defensive cybersecurity work. The tiers are structured by risk level: Defense Access supports general security operations, Red Team Access allows authorized penetration testing, and Specialized Access permits safety testing on critical infrastructure systems. To ensure safety, Anthropic conducted evaluations using CyScenarioBench, demonstrating that higher-tier access effectively removes blocks while lower tiers maintain significant safeguards against malicious use. The company reported that partners previously identified over 129,000 verified software vulnerabilities through earlier programs, highlighting the tool's impact on vulnerability discovery. Data retention is currently required for monitoring purposes, though a new Enterprise Frontier Safeguards solution offering zero data retention will be available later this fall. Eligible organizations can apply for access through the Claude Platform, Google Cloud’s Vertex AI, or Microsoft Foundry.
anthropic