Artificial intelligence company Anthropic has launched the Anthropic Cyber Mission, a long-term initiative aimed at helping cybersecurity defenders protect critical infrastructure and strengthen the security of open-source software.

The initiative will provide defenders with AI tools, security research, engineering expertise and other resources to identify and address vulnerabilities across essential systems.

The launch includes two programmes: the Critical Infrastructure Defense Program (CIDP), which focuses on operational technology and government systems, and OSS Scanner, a free service offering regular security scans for open-source software projects.

Anthropic said the initiative comes as advanced AI capabilities make it easier to identify software vulnerabilities, while organisations continue to face challenges in verifying findings, prioritising threats and implementing fixes.

State-sponsored adversaries have established footholds in systems across multiple sectors, raising concerns about the security of infrastructure that supports essential services.

New programme targets critical infrastructure security

The Critical Infrastructure Defense Program will provide selected security providers with access to Anthropic’s Claude AI models, on-site engineers and threat research to help protect operational technology (OT) environments.

These systems underpin essential services such as electricity generation and distribution, water utilities, manufacturing facilities and transportation networks.

Unlike conventional IT infrastructure, many OT systems are designed to operate for decades and cannot easily be taken offline for security updates. Proprietary equipment, operational constraints and the risks associated with system changes can leave known vulnerabilities unresolved for extended periods.

Anthropic said the programme will work with trusted providers to help operators identify weaknesses and determine appropriate remediation measures without unnecessarily disrupting critical operations.

The programme’s founding partners are Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC and Rockwell Automation.

The partners bring expertise spanning cybersecurity consulting, industrial network security, advanced security technologies and industrial equipment manufacturing.

Anthropic said several partners are already using Claude to address vulnerabilities and help customers implement fixes. The initial phase will involve a small group of providers to evaluate which approaches are most effective and practical.

Organisations developing security products or services for critical infrastructure can register their interest in joining the programme.

Free AI-powered security scans for open-source projects

Anthropic has also introduced OSS Scanner, which will provide open-source software projects with regular security scans using the company’s most capable AI models at no cost.

Open-source software forms part of the foundation of many applications and digital services. However, project maintainers often operate with limited funding and personnel, despite the widespread use of their code.

The scanner is intended to help maintainers identify vulnerabilities and support efforts to develop patches, strengthening the security of software used by developers, businesses and public-sector organisations.

The initiative focuses on improving the security of shared code that supports much of the wider technology ecosystem.

Initiative expands on Project Glasswing

The Anthropic Cyber Mission builds on work undertaken through Project Glasswing, where participating partners identified numerous vulnerabilities.

Anthropic said finding vulnerabilities is only one part of reducing cyber risk, with verification, prioritisation and remediation remaining significant challenges.

Earlier this week, the company merged Project Glasswing into its expanded Cyber Verification Program, which gives a wider group of defenders access to its most capable AI models.

Through the Cyber Mission, Anthropic plans to combine AI tools with engineering expertise, research and funding to support organisations responsible for securing critical infrastructure and open-source software.

The company also acknowledged the dual-use risks associated with advanced AI in cybersecurity. Although frontier models can help defenders discover and fix vulnerabilities, they may also be misused to exploit weaknesses and conduct malicious cyber operations.

Anthropic said it intends to expand the Cyber Mission over time, introducing additional tools, research and resources to support cybersecurity defenders in other areas.

The initial focus on critical infrastructure and open-source software reflects the company’s stated goal of strengthening the systems and shared code on which essential services and digital technologies depend.

If your company builds security products or services for critical infrastructure, you can register your interest here.

Source — Anthropic