Anthropic has merged its Cyber Verification Program and Project Glasswing into one program with three access tiers. The company says verified security teams get advanced model capabilities with fewer automated blocks on work such as malware analysis and vulnerability validation.

In its Oct. 6 announcement, Anthropic describes Defense Access as the tier for security operations and incident response, malware reverse engineering, and vulnerability analysis. Eligible applicants include corporate security teams, nonprofits, universities, government bodies, critical infrastructure operators of any size, smaller security firms, open-source maintainers and researchers with a record of reported vulnerabilities. Anthropic aims to answer applications within a few days. Red Team Access adds authorized penetration testing for organizations, takes a few weeks to review, and is closed to individuals. Specialized Access has the fewest blocks and is reserved for a limited set of verified organizations, each reviewed in depth with the US government.

Real-time blocks remain on actions that could cause physical harm or mass disruption, such as deploying ransomware. Enrolled organizations must allow data retention so Anthropic can monitor for misuse, and the company says a zero-data-retention option with safeguards is due later this fall.

Anthropic reports that Glasswing partners found at least 129,000 verified software vulnerabilities between April and July 2026, and that more than 33,000 verified findings so far are rated critical or high. It says the figures rest on survey data from a subset of partners, so the true total is likely higher.

Our read: for a SOC lead, the first questions are procurement ones. The data-retention requirement needs privacy and legal sign-off before an application is worth filing, and the tier should match the work the team does today. For other recent AI-agent incidents, see Wikimedia Reports AI Agent Edits, Etherpad Probes and Heavy API Traffic and DIVD Says Two Zammad Zero-Days Let an Apparent AI Agent Breach It.

Source: Anthropic, Expanding the Cyber Verification Program