Anthropic Unveils Project Glasswing: A Revolutionary Cybersecurity Initiative with a Restricted AI Model
AI pioneer, Anthropic, has recently launched Project Glasswing. This revolutionary cybersecurity initiative restricts access to their most potent AI model to date, in response to unprecedented security threats. On April 7, 2026, the company unveiled its new Claude Mythos Preview model, which boasts the capability to identify and exploit software vulnerabilities better than nearly any human security researcher.
In recent weeks, this AI model has autonomously unearthed thousands of zero-day vulnerabilities across all major operating systems and web browsers. This includes bugs that have remained undetected for up to 27 years. Instead of making the model publicly accessible, Anthropic has chosen to collaborate with tech behemoths such as Amazon Web Services, Apple, Google, Microsoft, NVIDIA, CrowdStrike, JPMorgan Chase, Cisco, and Palo Alto Networks in a pressing defensive maneuver.
Anthropic is pledging $100 million in usage credits and an additional $4 million in direct donations to open-source security organizations. According to project partners, “AI capabilities have crossed a threshold that fundamentally changes the urgency required to protect critical infrastructure from cyber threats.” The initiative’s primary goal is to provide defenders with a head start before similar AI capabilities become widely accessible to potential attackers.
Source: Anthropic Project Glasswing