Anthropic Launches Cyber Mission, Reveals 129,000+ Vulnerabilities Found

Anthropic unveiled a comprehensive cybersecurity commitment this week, positioning defensive security work as the company's clearest enterprise differentiator.

The new "Anthropic Cyber Mission," announced October 8, includes two tracks:

• Critical Infrastructure Defense Program — pairs Claude models and on-site engineers with providers securing power grids, water systems, and transportation networks

• Free OSS Scanner — an opt-in service giving open-source projects automated security reports from Anthropic's most capable models

Eleven organizations signed on as founding partners for the critical infrastructure track, including Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Palo Alto Networks, PwC, and Rockwell Automation.

Separately, Anthropic disclosed Project Glasswing results:

• 129,000+ previously unknown software vulnerabilities verified between April and July

• 5,500+ additional vulnerabilities found since through open-source scanning

• Over 33,000 rated critical or high severity

• The Cyber Verification Program expanded into three formal tiers — Defense, Red Team, and Specialized — letting vetted security professionals test Claude with reduced safety restrictions

By pairing a free public-good tool with revenue-generating enterprise programs, Anthropic is treating cybersecurity as a genuine commercial wedge rather than a side project.

Why It Matters: The 129,000-vulnerability figure gives the industry its first large-scale evidence that AI-assisted vulnerability hunting works at real scale, while the partner roster signals serious enterprise traction.