Jane Edwards 
- Anthropic has broadened access to its cyber-capable Claude models
- CVP now spans Defense, Red Team and Specialized Access levels
- The 2026 FedCiv Summit will explore OneGov, AI adoption and more
Anthropic has launched an expanded Cyber Verification Program, or CVP, that gives qualifying cybersecurity professionals access to its most advanced artificial intelligence models with fewer cyber-related restrictions.
As Anthropic widens access to its AI models for security teams and government users, federal technology leaders will gather at the Potomac Officers Club’s 2026 FedCiv Summit on Oct. 29 to discuss how agencies will invest in IT modernization. The event will feature discussions on the General Services Administration’s OneGov strategy, artificial intelligence adoption, cloud computing, AI-enabled IT modernization, zero trust and more. Sign up now to join the conversation.
The company said Tuesday the updated CVP folds its Project Glasswing initiative into a single program, with each of the three new access tiers covering Claude Opus 5.5, Claude Sonnet 5.5, Claude Mythos 5.1 and future models.
What Is CVP?
CVP is a trusted access program through which Anthropic verifies security professionals and allows them to use Claude models with fewer cyber restrictions than its generally available versions, whose safety classifiers may interrupt tasks such as malware analysis and exploit validation.
The updated program combines two efforts Anthropic ran over the past six months: the original CVP, which lowered safeguards on Claude Opus and Claude Sonnet models for verified teams with a single level of access, and Project Glasswing, which gave select organizations securing critical software access to Claude Mythos.
What Are the CVP Access Tiers?
Anthropic assigns access based on the type of cyber work an organization performs:
- Defense Access covers defensive work such as incident response, malware analysis and vulnerability validation. It is open to security teams, critical infrastructure operators, open-source maintainers and individual researchers.
- Red Team Access adds authorized penetration testing and red teaming for in-house and government red teams and penetration testing firms. Individual researchers are not eligible, and actions that could cause physical harm, such as deploying ransomware, remain blocked.
- Specialized Access has the fewest cyber blocks and is reserved for verified organizations authorized to test safety-critical systems such as power grids and telecom networks. Current Project Glasswing members will move into this tier.
Anthropic aims to review Defense Access applications within a few days, while Red Team Access reviews take a few weeks. Applicants are placed in Defense Access while they wait. Specialized Access applicants are vetted in collaboration with the U.S. government.
What Other Details Did Anthropic Share?
Anthropic reported that Project Glasswing partners identified at least 129,000 verified software vulnerabilities from April to July 2026, and its own open-source scanning surfaced an additional 5,500 from April to October. More than 33,000 of the verified findings have been rated critical or high severity.
CVP is available through the Claude Platform, Google Cloud’s Vertex AI and Microsoft Foundry. Enrolled organizations must maintain data retention so Anthropic can monitor for misuse until Enterprise Frontier Safeguards launches later this fall.
Current CVP members will retain their existing settings for earlier models and will be automatically reviewed for access to Claude Opus 5.5, Claude Sonnet 5.5 and Claude Mythos 5.1.
How Has Anthropic Expanded Government Access to Claude?
The CVP expansion follows several steps Anthropic has taken to widen government access to its AI models. In July, the company restored access to Claude Fable 5 and Mythos 5 after the U.S. lifted export controls on the models. It also committed to giving government partners early access to test models and safeguards before deployment.
In September, Nextgov/FCW reported that Anthropic had extended its $1-per-user OneGov offer for federal agencies through Oct. 31. Most recently, the company made Claude for Government generally available to federal and state agencies, which moved the FedRAMP High-authorized platform out of a public beta that began in July.
