Anthropic expands Project Glasswing to protect critical infrastructure from cyberattacks

Anthropic expanded Project Glasswing to about 200 organizations, adding roughly 150 new partners in more than 15 countries and pushing its AI security program deeper into the systems that keep power, water, healthcare and communications running. The company said each new organization had to meet security requirements before gaining access, a sign that the rollout is now moving from a small launch cohort into a wider, more sensitive layer of national infrastructure.

The expansion sharpened the policy stakes around a simple tradeoff: faster cyber defense in exchange for greater dependence on a private U.S. vendor. Anthropic has said a major cyberattack on partner organizations could affect more than 100 million people, a scale that places the program well beyond ordinary enterprise security. The company described Project Glasswing as an effort to secure critical software for the AI era, while also telling officials that its model capabilities have been part of discussions with the U.S. government.

Project Glasswing launched on April 7, 2026 with about 50 partners, including Amazon Web Services, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA and Palo Alto Networks. Anthropic said the program has already helped partners uncover more than 10,000 high- or critical-severity vulnerabilities since launch. To support the effort, the company committed up to $100 million in usage credits for Claude Mythos Preview and $4 million in donations to open-source security organizations.

The broader partner list also showed how quickly the program has moved beyond software vendors into the operators of essential services. Verizon joined Project Glasswing on May 15, 2026 and said it was testing Mythos Preview to strengthen network security. Operational technology providers and industry groups had previously pushed back after being left out of the first rollout, arguing that the initial design overlooked a heavily exposed slice of critical infrastructure, including energy, water and transportation systems.

American Water also met with the Office of the National Cyber Director to discuss Mythos and AI-cybersecurity threats, underscoring how the debate has moved into the policy arena in Washington. Anthropic’s expansion may give defenders an earlier warning against attacks, but it also deepens reliance on one company to help safeguard infrastructure that governments have long treated as a core public responsibility.