Anthropic Expands AI Cybersecurity Access Program

Anthropic is merging its Project Glasswing into the Cyber Verification Program (CVP), creating a tiered access system for advanced AI cyber capabilities. The expanded CVP will offer three levels of access, including Defense Access for defensive tasks, Red Team Access for authorized penetration testing, and Specialized Access for high-risk testing. The program aims to provide vetted security professionals with varying levels of safeguards on models like Opus, Sonnet, and Mythos, which have been used to uncover thousands of software vulnerabilities. However, experts caution that while the tiered model reduces misuse risks, it does not eliminate them entirely.