Anthropic opens three verified access tiers that relax cyber safeguards on Claude for vetted security professionals
Under its expanded Cyber Verification Program, Anthropic will let qualifying security professionals use advanced cyber capabilities with reduced blocking classifiers. Access comes in three tiers. Each tier includes Claude Opus 5.5, Claude Sonnet 5.5 and Claude Mythos 5.1, and Anthropic says new models will be added as they ship. The announcement also cites safeguard testing on CyScenarioBench. The change comes as banks and enterprises warn about rising AI-driven cyber risk. On the same day, JPMorgan CEO Jamie Dimon said those risks rose sharply after Mythos launched. It also follows a New York City Council hearing where the major labs declined to guarantee that their agents will always obey safeguards. For security startups building on Claude, verified tiers offer access to capabilities that ordinary API users cannot reach. Vetting, identity checks and usage logging are likely to become a standard cost of shipping offensive security products on frontier models.
