#AnthropicLaunchesGlasswingProgram

Anthropic Launches Project Glasswing: A Major AI Cybersecurity Initiative

On April 7, 2026, Anthropic announced Project Glasswing, a collaborative initiative uniting major tech companies to use advanced AI for proactive cybersecurity defense.

Core Initiative: Putting a Powerful AI Model to Work Defensively

The project is centered on Claude Mythos Preview, an unreleased AI model. In testing, Mythos Preview scored 83.1 percent on the CyberGym security benchmark, compared to 66.6 percent for Claude Opus 4.6. It has already identified thousands of zero-day vulnerabilities, including a 27-year-old bug in OpenBSD and a 16-year-old flaw in FFmpeg that had survived millions of automated tests.

Due to its potential for misuse, the model will not be released to the public. Access is strictly limited to partners for defensive work.

Key Partners and Industry Support

The launch group includes AWS, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks. Over 40 additional organizations maintaining critical infrastructure will also participate.

Microsoft noted the model showed substantial improvements on its security benchmarks, while Cisco stated the old ways of hardening systems are no longer sufficient.

Investment in Open Source Security

To support the broader ecosystem, Anthropic is committing up to 100 million dollars in usage credits and making 4 million dollars in direct donations to open-source security organizations, including Alpha-Omega, OpenSSF, and the Apache Software Foundation.

What's Next

Participating organizations will share findings where possible, and Anthropic plans to report publicly on fixed vulnerabilities within 90 days. The company is also in ongoing discussions with US government officials about the model's capabilities.