Anthropic has announced the Anthropic Cyber Mission, a long-term initiative focused on strengthening cybersecurity for critical infrastructure and open-source software. The American AI company said the program will provide tools, research, and resources to help defenders secure essential systems. The initial phase prioritizes operational technology that supports power grids, water systems, transportation networks, and government systems.
Critical Infrastructure Defense Program
The Critical Infrastructure Defense Program brings frontier Claude models, on-site engineers, and threat research to the trusted providers that secure operational technology. Founding partners include Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC, and Rockwell Automation. Anthropic said work is already underway with several partners to identify and remediate vulnerabilities in these environments.
In June 2026, Anthropic launched a cyber defense program for state, local, tribal, and territorial governments in the United States. Since then, the company has provided frontier Claude models and technical support to more than half of US states and several large public infrastructure operators. The effort has focused on code scanning, patching, incident response, red teaming, and other security workflows.
Operational technology often cannot be taken offline for patching, which leaves known vulnerabilities unresolved for years. The equipment is proprietary, changes are risky, and a mistake can take down an entire plant. The program aims to help operators and their providers close security gaps before adversaries can disrupt essential services.
Securing Open Source Software
Anthropic also launched OSS Scanner, a free opt-in service that gives enrolled open source projects periodic security scans from its most capable models. Each report includes a proof of concept, an explanation, and a suggested fix where one is available. The reports are model-generated without human review, which speeds delivery but may include inaccuracies such as incorrect severity ratings.
The service builds on lessons from Project Glasswing, which scanned hundreds of widely used open source projects and privately reported potential vulnerabilities. Anthropic expects a true-positive rate above 90% and will work to improve accuracy and fix quality over time. The company also provided funding to organizations including the Python Software Foundation, the Apache Software Foundation, and Linux Foundation initiatives Alpha-Omega and OpenSSF.
Outlook and Expansion
Anthropic said highly capable AI models are already available to attackers, while defensive tools have not reached enough defenders. The company forecasts that within two years AI will favor defense by catching bugs before release, building secure software from scratch, and actively defending systems. In the near term, however, it warned that declining exploitation costs could increase cyber risk while verification and patching remain slow.
Anthropic observed that in Project Glasswing, months often passed between vulnerability discovery and remediation, and operational technology fixes can wait years or decades. The company plans to share research, tools, and resources, including unsuccessful experiments, with the broader security community. It also expressed intent to collaborate with other AI developers, security companies, and governments to strengthen collective defense.
The Anthropic Cyber Mission represents a structured effort to shift security advantages toward defenders of critical infrastructure and open source software. By combining frontier models with on-site expertise and free scanning tools, the company aims to reduce exploitable pathways and accelerate response. Anthropic invited security vendors, system integrators, equipment manufacturers, open source maintainers, and security teams to register interest or apply for access as the initiative expands.