Anthropic Launches Project Glasswing to Secure Critical Software with AI

Anthropic has launched Project Glasswing, a cross-industry initiative aimed at strengthening global cybersecurity using advanced AI capabilities. The project brings together partners including Amazon Web Services, Apple Inc., Broadcom Inc., Cisco, CrowdStrike, Google, JPMorgan Chase & Co, the Linux Foundation, Microsoft Healthcare, NVIDIA, and Palo Alto Networks.

At the center of the initiative is the unreleased Claude Mythos model, which has demonstrated the ability to autonomously identify and exploit high-severity vulnerabilities in major operating systems, browsers, and other critical software. The model has already uncovered thousands of previously unknown security flaws, some existing for decades.

Project Glasswing partners will use Claude Mythos to detect and patch vulnerabilities in their own systems and open-source software. Anthropic is extending access to more than 40 additional organizations responsible for key software infrastructure. The company has committed up to $100 million in model usage credits and $4 million in direct donations to open-source security groups, including the Linux Foundation’s Alpha-Omega and OpenSSF initiatives and the Apache Software Foundation.

According to Anthropic, the project aims to apply frontier AI capabilities defensively before similar tools can be misused. Within 90 days, the company plans to publish findings on vulnerabilities addressed and best practices developed. Discussions are also underway with U.S. government agencies to assess the national security implications of these AI systems.

Project Glasswing marks the beginning of a long-term effort to redefine cybersecurity standards in the AI era, with participants expected to collaborate on updated approaches to vulnerability disclosure, software updates, and secure development practices.