Fleet Joins Anthropic's Project Glasswing to Test AI-Driven Endpoint Security

Anthropic Project Glasswing has a new security research partner. Fleet Device Management - an open-source endpoint governance platform managing millions of devices across 90 countries - joined the program this week, gaining early access to Claude Mythos Preview to test how frontier AI can close the gap between when a vulnerability surfaces and when a patch deploys.

What Fleet Brings to the Program

Fleet handles device management for Macs, PCs, iPhones, Android devices, Chromebooks, desktop Linux, robotics, and cloud infrastructure - all through a single open-source platform. That is a large attack surface to monitor. Inside the Glasswing program, the team will evaluate AI-assisted workflows for understanding endpoint state, ranking what needs patching first, and speeding up internal security operations when active threats emerge.

The 79% Patching Gap Fleet Is Targeting

Research behind the collaboration names a specific problem: 79% of organizations take more than 24 hours to deploy critical security patches after release. Attackers now exploit most publicly disclosed vulnerabilities within hours. That is the operational gap AI security tooling keeps promising to close - and where most of those promises are still unproven in production. The same dynamic drove the Linux Foundation to build Akrites, an AI-led effort to push open-source patch velocity past its current floor.

Open-Source Research With Public Findings

Fleet will publish what it learns through the program, sharing findings and best practices with the broader open-source security community rather than keeping results internal. Anthropic Project Glasswing grants partners controlled access to Claude Mythos Preview; what gets built inside the program cannot be redistributed, but what gets learned can go public. That distinction matters for an open-source platform with a community depending on the output.

As a company that specializes in making IT and security teams' jobs easier, celebrating Fleet's founding anniversary with a renewed commitment to our own product security feels fitting.

Mike McNeil, CEO, Fleet Device Management

Fleet turns six today. Founded September 15, 2020, the company timed this week's Anthropic Project Glasswing announcement to the anniversary - framing it as a security commitment rather than a standard partnership note.

Where Glasswing Fits in Anthropic's Security Work

Project Glasswing is Anthropic's controlled-access program for evaluating how its frontier models strengthen defensive security research. Anthropic has been building out that security posture since publishing Fable 5's cybersecurity rulebook and a severity scale for AI jailbreak attempts - framing how it expects partners and researchers to engage with the models in high-risk domains. Fleet is now part of that network.

No public roster exists for Glasswing. Anthropic has not disclosed how many organizations currently hold access to Claude Mythos Preview through the program, or whether the admission process will open to applicants outside the current cohort.