Anthropic announced the Anthropic Cyber Mission on Thursday: a Critical Infrastructure Defense Program pairing its models with on-site engineers, backed by partners including Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Nozomi Networks, Palo Alto Networks, PwC and Rockwell Automation, alongside an OSS Scanner offering free security scans from its strongest models to established open-source projects with critical infrastructure impact. The company said attackers will keep the advantage for the next two years before defenders pull ahead, and that LLMs now find more than 85 percent of vulnerabilities, up from 20 percent at the start of 2025. The push follows a year in which frontier-model agents broke into at least 20 organizations, most notably Hugging Face, drawing government backlash in Washington.
Anthropic launches cyber mission for critical infrastructure
Anthropic launched the Anthropic Cyber Mission: a critical-infrastructure defense program pairing its models with on-site engineers, plus free security scans for critical open-source projects.
Why it matters: A frontier lab is putting its own models on the defensive side of AI-discovered zero-days after a year in which AI agents broke into dozens of organizations, per the report.
Data as of October 8-9, 2026. Single-sourced: The Register reporting Anthropic's blog post and launch announcement. Partner list, the two-year attacker-advantage timeline, and the 85%-vs-20% vulnerability-finding figures are Anthropic's claims, not independently verified.
Sources
- AI company moves to defend critical infrastructure and open-source projects from AI · The Register · 2026-10-08