Anthropic said it is launching an expanded version of its Cyber Verification Program, which makes advanced cyber capabilities and reduced blocking classifiers available to qualifying security professionals. The programme now has three access tiers: Defense Access, for defensive work such as security operations, incident response, malware reverse-engineering and vulnerability analysis; Red Team Access, which adds authorised penetration testing and red-teaming; and Specialized Access, reserved for a limited set of verified organizations authorised to test safety-critical systems such as flight operating systems, power grids, telecom networks, interbank transfer infrastructure and government administrative networks. Anthropic says each tier includes access to its most capable models, including Claude Opus 5.5, Claude Sonnet 5.5 and Claude Mythos 5.1, and that it is folding two earlier programmes, Project Glasswing and the original CVP, into one offering.
Our takeThis is a sensible answer to a real dilemma: cyber capability is dual-use, so the choice is not to block it or ship it but to vet who gets it. Tiering by scope of work, with real-time blocks still on for mass-disruption actions such as ransomware, is more honest than a one-size classifier. The catch is the verification pipeline itself, with weeks-long reviews, organisations-only red-teaming and a judgement call on every applicant. That does not scale cheaply, and the frontier labs are all quietly building the same gatekeeper role. Worth watching as a template, not a solved problem.