Anthropic today announced the Anthropic Cyber Mission, a long-term commitment to securing the systems everyone depends on, with tools, research, and resources for defenders. The effort starts with two areas: critical infrastructure and open-source software.

For critical infrastructure, the new Critical Infrastructure Defense Program (CIDP) brings frontier Claude models, on-site engineers, and threat research to the trusted providers that secure operational technology — the controllers, control software, and industrial networks behind power grids, water utilities, factories, and transportation. These systems often cannot be taken offline to patch, so known vulnerabilities can persist for years; equipment is proprietary, changes are risky, and a mistake can shut down a plant. Founding partners are Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC, and Rockwell Automation. Several partners are already working with Claude to fix vulnerabilities, and Anthropic says its first step is a small cohort of providers to learn which strategies are most effective and practical. Partner statements frame operational technology as the next frontier for AI-enabled attacks and argue that machine-speed threats require machine-speed defense.

For open-source software, Anthropic launched OSS Scanner, an opt-in, free service inspired by Google’s OSS-Fuzz. Enrolled projects receive periodic scans from Anthropic’s most capable models. Each report includes a proof of concept of how the bug could be exploited, an explanation, and a suggested fix where one is available. Reports are model-generated and sent without human review, so maintainers get them faster but some will contain inaccuracies such as wrong severity ratings; Anthropic expects a true-positive rate above 90% and plans to improve it. The service is intended for projects with the capacity to keep up with surfaced findings; for others, Anthropic will continue to share human-verified disclosures under its coordinated vulnerability disclosure policy.

Both efforts build on lessons from Project Glasswing, under which Anthropic scanned hundreds of widely used open-source projects, had humans triage and review many findings, and privately reported them to maintainers. The key lesson: finding vulnerabilities is easier than ever, but verifying, prioritizing, and fixing them remains hard. Earlier this week, Project Glasswing was merged into the expanded Cyber Verification Program, which gives more defenders access to Anthropic’s most capable models. In June, Anthropic also launched a cyber defense program for state, local, tribal, and territorial governments, and has since offered frontier Claude models and technical support to more than half of all US states and some of the country’s largest public critical infrastructure operators.

Beyond OSS Scanner, Anthropic’s open-source work will aim to get findings to maintainers faster, accelerate fixes by automating triage and patching, and explore new secure architectures and coding practices. Anthropic has funded organizations behind widely used open-source code, including the Python Software Foundation, Alpha-Omega and OpenSSF through the Linux Foundation, and the Apache Software Foundation, plus Akrites and Gold Eagle, which collect and coordinate vulnerability reports. The Defender Advantage Fund (0xDAF), launched in August, supports pilot programs and keeps OSS Scanner free. Maintainers can also apply to Claude for Open Source for free Claude Max subscriptions and to the Cyber Verification Program for expanded cyber capabilities for defensive work.

The rationale: highly cyber-capable AI models are widely available to attackers, but defensive tools — including Anthropic’s own — have not reached enough defenders. Anthropic’s forecast is that in two years AI will favor defense: easier to catch bugs before they ship, write fundamentally secure software from scratch, and actively defend systems. In the near term, that may not be true: the cost of exploiting vulnerabilities has dropped, while verifying, disclosing, and fixing them is slow and still depends on people. In Glasswing, months often passed between finding and fixing a vulnerability, and operational technology fixes can wait years, in rare cases decades, until they can be applied safely to running machinery. Success means critical infrastructure such as water, power, transport, and communications keeps running under attack by capable adversaries, with fewer exploitable paths and faster recovery when something gets through.

Over the coming months, Anthropic will expand CIDP to more partners and sectors and share what it learns, including what did not work. It also expects other AI developers, security companies, and governments to run their own efforts and hopes to collaborate with them. Companies that build security products or services for critical infrastructure can register interest in CIDP; core maintainers of critical open-source projects can enroll in OSS Scanner; and security teams can apply to the expanded Cyber Verification Program for access to the models.