Anthropic Expands Cyber Verification Program Into Three Access Tiers | Security
Anthropic announced on October 6, 2026 an expanded Cyber Verification Program (CVP) that gives vetted security professionals reduced blocking cyber classifiers…
Published on MyPrivateClaw
Oct 8, 2026, 11:04 PM UTC
Coverage date
Oct 7, 2026
Last updated
Oct 8, 2026, 11:04 PM UTC
News summary
Anthropic announced on October 6, 2026 an expanded Cyber Verification Program (CVP) that gives vetted security professionals reduced blocking cyber classifiers and access to its most capable models, integrating its earlier Project Glasswing and CVP efforts into one offering with three access tiers: Defense Access, Red Team Access, and Specialized Access. Each expanded CVP access tier provides access to the named AI models Claude Opus 5.5, Claude Sonnet 5.5, and Claude Mythos 5.1, plus new models moving forward, with existing CVP and Glasswing members kept on their current settings and auto evaluated for access to those models. Anthropic says its generally available models, including Claude Opus 5.5, Claude Fable 5.1, and Claude Sonnet 5.5, carry conservative cyber safeguards that block most cyber work, and the expanded program lowers those blocks for qualifying defenders across Defense,…