Anthropic launches Critical Infrastructure Defense Program with 11 security partners

Anthropic launched the Critical Infrastructure Defense Program on October 8, 2026, aiming to secure operational technology environments. This initiative is part of the broader Anthropic Cyber Mission, which focuses on protecting critical infrastructure and open-source software through artificial intelligence tools and technical resources.

The program combines Anthropic’s frontier Claude models with on-site engineering support and threat research. It targets operational technology systems that support power grids, water treatment facilities, factories, and transportation networks. The effort seeks to help defenders find and repair weaknesses before adversaries can exploit them to disrupt essential services.

Strategic Partnerships and Execution

Eleven organizations serve as founding partners in the program. The group includes Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC, and Rockwell Automation. Anthropic described these partners as providers that operators rely on to secure systems.

Anthropic stated that several partners are already working with Claude to address vulnerabilities. These providers help customers fix issues by assessing which approaches are most effective and practical. The company emphasized that this work is underway, with a small cohort of providers testing strategies to learn what works safely.

Challenges in Operational Technology

Critical infrastructure systems often blend modern information technology with older operational technology. These industrial systems can remain in service for decades, creating complex security environments. Taking them offline for patching may interrupt essential services, leading operators to leave known weaknesses unresolved for years.

Anthropic noted that state-sponsored adversaries have spent years gaining footholds in these systems. The increasing availability of highly capable AI models has reduced the cost of exploiting vulnerabilities. This dynamic makes proactive defense urgent, as legacy systems may retain known issues due to operational constraints, proprietary equipment, and safety risks.

What should you do?

Security professionals should monitor announcements from the founding partners for updates on implementation. Organizations using legacy OT systems should review their current patching constraints and exposure. Teams should assess how AI tools might integrate with existing security workflows without disrupting operations.

What to do and how to stay safe: Anthropic

• Review current operational technology assets for known vulnerabilities that cannot be patched immediately.
• Monitor communications from Anthropic and its 11 founding partners for available defensive tools.
• Evaluate how AI-driven threat research might complement existing security monitoring programs.
• Ensure security teams understand the specific risks posed by legacy equipment in their environment.

Step-by-step guide: Prompt Injection Warning Signs: How to Spot AI Manipulation

General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

When did Anthropic launch the Critical Infrastructure Defense Program?

Anthropic announced the launch of the Critical Infrastructure Defense Program on October 8, 2026.

Which companies are founding partners in the program?

Founding partners include Accenture, Booz Allen, CrowdStrike, Deloitte, Dragos, Hitachi, Insane Cyber, Nozomi Networks, Palo Alto Networks, PwC, and Rockwell Automation.

How does CrowdStrike contribute to the Anthropic initiative?

CrowdStrike contributes security data, threat intelligence, and experience from its Falcon platform, including Falcon for XIoT, to help discover unmanaged devices and detect threats.