Anthropic Launches Project Glasswing Cybersecurity Initiative with New Mythos AI Model

Apr 7, 2026

Anthropic released a preview of Mythos, a powerful new AI model, on April 7, 2026, as part of Project Glasswing, a cybersecurity partnership with over 40 organizations including Nvidia, Google, AWS, Apple, and Microsoft. The model, which has strong coding and reasoning capabilities, identified thousands of zero-day vulnerabilities in recent weeks, many critical and one to two decades old, and will be used by partners to scan software systems for security flaws. The launch follows a turbulent period for Anthropic, including a March 2026 data leak that exposed the project (originally codenamed Capybara) and accidentally revealed nearly 2,000 source code files. Anthropic faces ongoing tensions with the Trump administration through a legal battle and Pentagon designation as a supply-chain risk due to the company's refusal to allow autonomous targeting or surveillance of U.S. citizens.

Launch of Mythos AI Model and Project Glasswing Initiative

▪Anthropic's Mythos model preview is part of a new security initiative called Project Glasswing.
▪Project Glasswing partners will share what they learned from using Anthropic's Mythos model so the rest of the tech industry can benefit.
▪More than 40 partner organizations will deploy Anthropic's Mythos model for defensive security work and to secure critical software.
▪Partner organizations previewing Anthropic's Mythos include Amazon, Apple, Broadcom, Cisco, CrowdStrike, the Linux Foundation, Microsoft, and Palo Alto Networks.
▪Anthropic released a preview of its new frontier model Mythos on Tuesday, April 7, 2026.
▪Anthropic's Mythos model preview will not be made generally available.

Previous Data Leaks and Security Incidents at Anthropic

▪News of Anthropic's Mythos model was originally leaked in a data security incident reported in March 2026 by Fortune.
▪Anthropic accidentally exposed nearly 2,000 source code files and over half a million lines of code in March 2026 via a mistake in the launch of version 2.1.88 of its Claude Code software package.
▪Anthropic accidentally caused thousands of code repositories on Github to be taken down while attempting to clean up the Claude Code source code exposure.

Legal and Regulatory Complications with Federal Government

▪The Pentagon labeled Anthropic a supply-chain risk over Anthropic's refusal to allow autonomous targeting or surveillance of U.S. citizens.

Perspective of Anthropic

▪Anthropic designed Project Glasswing so partner organizations can share learnings to benefit the broader tech industry.
▪Anthropic attributed the March 2026 Mythos data leak to human error.

Perspective of Pentagon and Trump administration

▪The Pentagon labeled Anthropic a supply-chain risk over restrictions on autonomous targeting and surveillance capabilities.
▪The Trump administration is currently locked in a legal battle with Anthropic.

Perspective of Project Glasswing partner organizations

▪More than 40 partner organizations will deploy Anthropic's Mythos model for defensive security work and to secure critical software.
▪Project Glasswing partners will use Mythos to scan both first-party and open-source software systems for code vulnerabilities.