In early October, Anthropic announced the Anthropic Cyber Mission, a long-term cyber defense initiative that aims to harness Claude's capabilities for defenders. According to reports, it launched with two areas: a critical infrastructure defense program, and a free AI-based vulnerability scanning service for open-source projects.
The partners
The initiative launched with 11 founding partners, including, according to a report in Japan, Accenture, CrowdStrike, Dragos, which specializes in industrial systems security, and Palo Alto Networks. Critical infrastructure, such as power, water and communications networks, is a favored target for state attackers, and defending it requires cooperation between companies.
Free open source scanning
Almost all software relies on open-source libraries, often maintained by volunteers with no security budget. Anthropic's service lets core maintainers sign up and get their code scanned. The initiative builds on Project Glasswing, which the company began in April, and Anthropic plans to expand it with tools, research and funding.
The criticism
According to one analysis, the scan reports include proof-of-concept exploit code generated without human review before reaching maintainers. It again raises the dilemma of AI in cybersecurity: a tool that finds and demonstrates flaws helps defenders, but demands caution. Google, with Gemini 4 Carbon, and OpenAI take a similar "defenders first" approach.
