Anthropic Unveils Project Glasswing to Combat AI Cyberattacks

Rowan Ashford

Published: April 8, 2026 1:36 AM ET

Anthropic Unveils Project Glasswing to Combat AI Cyberattacks

Concerns about the negative impact of artificial intelligence, particularly in cybersecurity, have intensified. In response, Anthropic has launched an initiative known as Project Glasswing, aimed at fortifying critical software against AI-driven cyber threats.

Overview of Project Glasswing

Project Glasswing involves collaboration with prominent technology firms and organizations. The initiative is designed to enhance the security landscape by addressing vulnerabilities associated with AI.

Utilization of Claude Mythos Preview

Participants in Project Glasswing will leverage Claude Mythos Preview, an unreleased general-purpose model from Anthropic. This model has already identified thousands of exploitable vulnerabilities, impacting various major operating systems and web browsers.

AI Defense for Cybersecurity

Anthropic aims to deploy its tools defensively. The goal is to mitigate the malicious use of AI, which poses serious risks to both economic stability and national security.

Ethical Concerns and Challenges

Anthropic has positioned itself as a key player in addressing ethical issues surrounding artificial intelligence. Earlier this year, the company faced challenges when it declined to remove safety protocols from its services destined for the Pentagon. This decision led to a supply chain risk designation by the Department of Defense.

Addressing Past Incidents

While Project Glasswing represents a proactive approach to cybersecurity, past incidents highlight ongoing challenges. In February, a hacker utilized Anthropic’s Claude against several government agencies in Mexico.

Launching Project Glasswing could represent a significant step towards enhancing cybersecurity in the AI era, despite the complexities of existing threats.