search
Project Glasswing
Trends
- 1Anthropic opens Claude to vetted cybersecurity experts●Discover how the Anthropic Cyber Verification Program grants cybersecurity experts access to Claude models to find and f
Anthropic is running a Cyber Verification Program that gives approved security researchers expanded access to its Claude AI models so they can hunt for and fix critical software vulnerabilities. The initiative, discussed alongside projects like Glasswing, is being shared in cybersecurity circles as an example of AI companies channelling frontier models toward defensive security work rather than restricting them.
- 2Anthropic Expands Cyber Program Giving Vetted Defenders Reduced-Safeguard Claude Access●Anthropic expanded its Cyber Verification Program, giving vetted defenders access to Claude models with reduced safeguar
Anthropic has expanded its Cyber Verification Program, allowing vetted cybersecurity defenders to use Claude models with reduced safeguards for defensive work. The company's Project Glasswing reportedly identified 129,000 verified vulnerabilities between April and July 2026, plus 5,500 additional findings, some of which are now being actively exploited. Security professionals are debating the balance between giving defenders powerful AI tools and the risks of weakened model safeguards.