Anthropic is launching a new AI model for cybersecurity
Anthropic is introducing Project Glasswing and the Claude Mythos Preview model to help organizations autonomously detect system vulnerabilities.Summary
Anthropic has announced Project Glasswing, a cybersecurity initiative in partnership with major tech companies including Nvidia, Google, Amazon Web Services, Apple, and Microsoft. Central to this project is Claude Mythos Preview, a new general-purpose model designed to assist cyber defenders by autonomously identifying and patching high-severity vulnerabilities in operating systems and web browsers without human intervention.
To prevent adversaries from exploiting the model's capabilities, Anthropic is restricting access to a select group of partners, which includes JPMorgan Chase, Cisco, and CrowdStrike, among others. While the model was not specifically trained solely for cybersecurity, its advanced reasoning and agentic coding skills enable it to function effectively in this domain.
Anthropic is also engaging in discussions with US government officials regarding the model's offensive and defensive capabilities. To support the ecosystem, the company is committing up to $100 million in usage credits and making direct donations to the Linux Foundation and the Apache Software Foundation.
(Source：The Verge)