Anthropic is expanding a program that allows vetted cybersecurity specialists to test its most powerful artificial intelligence models with fewer safeguards. This follows the initiative Project Glasswing, which helped uncover over one hundred thousand software vulnerabilities this year.
Partners within Glasswing, a program aimed at securing the world's most critical software, discovered at least 129,000 confirmed vulnerabilities between April and July. Additionally, Anthropic's proprietary open-source scanning identified another 5,500 vulnerabilities between April and October.
More than 33,000 of these vulnerabilities were classified as critical or high severity. Anthropic notes that these figures are likely underestimated and expects the actual damage to be at least five times higher, as the data comes from surveying a limited number of partners.
The updated Cyber Verification Program (CVP), announced on Tuesday, merges two programs Anthropic has run over the past six months. The first, Glasswing, provided organizations involved in securing critical software with access to Claude Mythos—Anthropic's most cyber-resilient model line. The second, the original CVP, offered vetted security teams reduced restrictions on Claude Opus and Sonnet models.
Previously, the introduction of Claude Mythos Preview in April raised concerns that AI could hack software before it was secured. The new program includes three tiers, each with its own verification requirements and security controls. All three tiers provide access to Claude Opus 5.5, Sonnet 5.5, Mythos 5.1, and future models.
The Defense tier is intended for tasks such as malware analysis and incident response. Security teams, critical infrastructure operators, open-source developers, and vulnerability reporting researchers can apply. The Red Team tier involves authorized penetration testing and 'red team' exercises, but only organizations can apply. The Specialized tier has the fewest restrictions and is reserved for a small group of organizations authorized to test security-critical systems, such as power grids, aviation systems, and interbank transfer infrastructure. Anthropic conducts vetting of every participant jointly with the U.S. government, and existing Glasswing participants transition to this tier.