#AnthropicLaunchesGlasswingProgram Project Glasswing” with Tech Giants to Secure Critical Infrastructure

The AI firm unites Apple, Microsoft, Google, and Amazon in an urgent bid to find and fix vulnerabilities before malicious actors can exploit them.

SAN FRANCISCO – April 8, 2026 – Anthropic has announced Project Glasswing, a major new industry initiative designed to leverage advanced AI for defensive cybersecurity. The project brings together a coalition of technology and security partners—including Amazon, Apple, Cisco, CrowdStrike, Google, Microsoft, and Palo Alto Networks—to gain access to a powerful new, unreleased AI model: Claude Mythos Preview .

The launch comes in response to a critical inflection point in software security. Anthropic reports that AI models have reached a level of coding capability where they can surpass all but the most skilled humans at finding and exploiting software vulnerabilities. In recent weeks, the company states that Mythos Preview has already identified thousands of zero-day vulnerabilities, including a bug in OpenBSD that remained hidden for 27 years and a 16-year-old flaw in popular video software that eluded automated testing tools for millions of runs .

“AI capabilities have crossed a threshold that fundamentally changes the urgency required to protect critical infrastructure from cyber threats, and there is no going back,” said Anthony Grieco, Senior Vice President & Chief Security & Trust Officer at Cisco. “Our foundational work with these models has shown we can identify and fix security vulnerabilities at a pace and scale previously impossible” .

A Defensive “Urgency”

Unlike traditional security tools that rely on known signatures, Mythos Preview is a general-purpose model whose cyber performance comes from advanced reasoning and agentic coding rather than specific training. However, this power poses a dual risk. Anthropic is keeping the model under tight wraps, acknowledging that the same capabilities that find bugs could also be used to launch sophisticated attacks.

Project Glasswing is designed to get ahead of this curve. By granting access exclusively to vetted defenders, Anthropic aims to patch critical systems before these AI capabilities proliferate to malicious actors.

“As we enter a phase where cybersecurity is no longer bound by purely human capacity, the opportunity to use AI responsibly to improve security and risk at scale is unprecedented,” said Igor Tsyganskiy, Global CISO at Microsoft. “Joining Project Glasswing allows us to identify and mitigate risk early” .

Industry and Open Source Support

The coalition includes a wide swath of the tech economy, from cloud providers like AWS to hardware manufacturers like Broadcom and Nvidia, as well as financial giant JPMorganChase and the Linux Foundation .

Recognizing that open-source software constitutes the vast majority of modern code, Anthropic is extending access to roughly 40 additional organizations that maintain critical infrastructure. To facilitate this, the company has committed up to $100 million in usage credits** and is making **$4 million in direct donations to open-source security groups like the Apache Software Foundation and OpenSSF .

“Open source maintainers have historically been left to figure out security on their own,” said Jim Zemlin, CEO of the Linux Foundation. “By giving these maintainers access to a new generation of AI models, Project Glasswing offers a credible path to changing that equation” .

The Road Ahead

While the Claude Mythos Preview will not be released to the general public, participating organizations are already using it to scan their own systems. AWS noted that even in "well-tested environments," the model identified new opportunities to harden code . Palo Alto Networks added that the model found vulnerabilities that "prior-generation models missed entirely" .