Anthropic launched the Anthropic Cyber Mission on October 8, with two initial programs. The Critical Infrastructure Defense Program brings Claude models, on-site engineers and threat research to providers that secure industrial systems. OSS Scanner offers free, periodic security scans to open-source projects that opt in.
The initiative follows a limitation Anthropic saw in Project Glasswing. Its partners found vulnerabilities, but reviewing and fixing them still took time. Anthropic says months often passed between discovery and a fix. The Cyber Mission adds engineering support and funding to the company’s efforts to give defenders access to capable models.
Industrial security through existing providers
The Critical Infrastructure Defense Program starts with the operational technology behind power grids, water systems and transportation networks. These controllers and industrial networks can run for decades. Operators often cannot take them offline to patch, and a faulty change can stop a plant.
The program’s 11 founding partners are Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC and Rockwell Automation. Anthropic says several already use Claude to fix vulnerabilities and help customers do the same. It plans to expand the initial group as it learns which approaches work.
This extends earlier government work. Anthropic says that since launching a government cyber defense program in June, it has offered models and technical support to more than half of US states and some large public infrastructure operators.
Free scans move review work to maintainers
OSS Scanner gives enrolled projects reports with an exploit proof of concept, an explanation and a suggested fix where available. Anthropic sends the model-generated reports without human review. It expects a true-positive rate above 90%, while warning that reports can contain errors such as incorrect severity ratings.
That delivery model responds to maintainers who asked for all findings, including unreviewed ones. Anthropic says the service suits projects that can keep up with the reports. It will continue human-verified disclosures for projects that need them.
The Defender Advantage Fund, launched in August, supports pilot programs and keeps OSS Scanner free. Future work includes automating triage and patching for willing projects, plus research on safer code and architectures.
How it fits with existing cyber programs
Anthropic explicitly names Google’s OSS-Fuzz as the inspiration for OSS Scanner. The connection places its new service alongside an existing Google effort to support open-source security. The announcement describes Anthropic’s model-generated reports, but supplies no head-to-head results against OSS-Fuzz. It supports a comparison of purpose rather than detection performance.
Within Anthropic, the programs give defenders different kinds of help:
| Program | What the announcement offers | Intended participants |
|---|---|---|
| Critical Infrastructure Defense Program | Models, on-site engineers and threat research | Providers of infrastructure security products and services |
| OSS Scanner | Free periodic scans and unreviewed reports | Open-source projects with capacity to review findings |
| Cyber Verification Program | Expanded access to advanced cyber capabilities | Qualifying security teams, including operators and maintainers |
Anthropic merged Project Glasswing into the expanded Cyber Verification Program earlier in the launch week. The Cyber Mission adds support for deploying and acting on model findings. Anthropic also invites collaboration with other AI developers, security companies and governments as the work expands.
