What happened
On 8 October 2026, Anthropic published “Introducing the Anthropic Cyber Mission”. It opens: “Today we’re launching the Anthropic Cyber Mission, a long-term commitment to securing the systems everyone depends on.” Anthropic describes it as an effort to give defenders tools, research and resources, and says it starts in two areas: critical infrastructure and open-source software.
Key details
All figures and expectations below are Anthropic’s own.
- Critical Infrastructure Defense Program (CIDP): aimed at the operational technology behind power grids, water systems and transportation networks, plus government systems. Anthropic offers frontier Claude models, on-site engineers and threat research to providers that help operators secure those systems. The 11 founding partners are Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC and Rockwell Automation. Anthropic says it is starting with a small group of providers and has a form for companies to register interest.
- OSS Scanner: “an opt-in service inspired by Google’s OSS-Fuzz”. Enrolled projects “receive periodic scans from our most capable models, free of charge”. Reports include a proof of concept, an explanation and, where one exists, a suggested fix.
- Caveats Anthropic states: “The reports are model-generated and sent without human review.” It expects “a true-positive rate above 90%” and says it will work to improve it.
- Funding and support: Anthropic says it funded the Python Software Foundation, Alpha-Omega and OpenSSF (through the Linux Foundation) and the Apache Software Foundation, and that its Defender Advantage Fund, launched in August, keeps OSS Scanner free. Maintainers can apply to Claude for Open Source for free Claude Max subscriptions.
- Related: Anthropic says it merged Project Glasswing into its expanded Cyber Verification Program, which it announced on 6 October.
Why it matters
For open-source maintainers, the offer is free scanning, but with a trade-off Anthropic states itself: reports arrive without human review, so each one needs checking before anyone acts on it. A claimed true-positive rate above 90% is an expectation, not a measured result, and no independent evaluation was published with the announcement.
For companies that secure industrial systems, the program is a route to Anthropic’s models and engineers, limited for now to a small group. Anthropic also says some fixes in operational technology may take decades, a reminder that finding vulnerabilities is not the same as patching them.
Disclosure: Claude, made by Anthropic, was one of the AI tools used to research and draft this article, and Anthropic is the company whose program this article covers. Deyron Labs is independent of Anthropic. Claims about the program are Anthropic’s own unless stated otherwise.
Sources
- Introducing the Anthropic Cyber Mission Primary source , Anthropic