Project Glass Wing
An AI-driven cybersecurity coalition of 40 major companies, established by Anthropic, to find and fix software vulnerabilities over a 100-day period.
First Mentioned
4/26/2026, 3:51:49 AM
Last Updated
4/26/2026, 3:59:49 AM
Research Retrieved
4/26/2026, 3:59:49 AM
Summary
Project Glass Wing is a 100-day cybersecurity coalition established by Anthropic to secure foundational software infrastructure against vulnerabilities discovered by its advanced AI model, Mythos (Claude Mythos Preview). The initiative was formed after Mythos autonomously identified critical, decades-old security flaws in the Linux kernel, OpenBSD (a 27-year-old vulnerability), and FFmpeg (a 16-year-old vulnerability). To address these risks, Anthropic partnered with industry leaders including Amazon Web Services, Apple, Google, Microsoft, NVIDIA, and the Linux Foundation. The project aims to provide software maintainers with AI-powered tools to identify and patch vulnerabilities at scale, prioritizing a defense-first approach. Due to the model's significant capability for exploit discovery, Anthropic CEO Dario Amodei decided to withhold Mythos from public release, signaling a cautious approach to AI development as the industry nears Artificial General Intelligence (AGI).
Referenced in 1 Document
Research Data
Extracted Attributes
Founder
Anthropic
Duration
100 days (Coalition period)
Lead Executive
Dario Amodei
Testing Metric
5 million automated tests failed to find the FFmpeg bug
Core Technology
Mythos (Claude Mythos Preview)
Target Software
Linux kernel, OpenBSD, FFmpeg
Vulnerabilities Identified
27-year-old flaw in OpenBSD, 16-year-old flaw in FFmpeg
Timeline
- Mythos AI autonomously discovers a 27-year-old vulnerability in OpenBSD and a 16-year-old vulnerability in FFmpeg. (Source: fccfeb90-332d-499c-b8d7-0026207a2740)
2024-01-01
- Anthropic establishes Project Glass Wing, a 100-day cybersecurity coalition with industry partners to patch foundational software flaws. (Source: fccfeb90-332d-499c-b8d7-0026207a2740)
2024-01-01
- Anthropic CEO Dario Amodei announces the decision to withhold the Mythos model from public release due to its potential for misuse in discovering zero-day exploits. (Source: fccfeb90-332d-499c-b8d7-0026207a2740)
2024-01-01
Wikipedia
View on WikipediaGreta oto
Greta oto is a species of brush-footed butterfly and member of the subfamily Danainae, tribe Ithomiini, and subtribe Godyridina. It is known by the common name glasswing butterfly for its transparent wings, which allow it to camouflage without extensive coloration. In Spanish-speaking regions, it may also be referred to as espejitos, meaning "little mirrors" because of its transparent wings. The butterfly is mainly found in Central America and northern regions of South America, with sightings as far north as Texas and as far south as Chile. While its wings appear delicate, the butterfly is able to carry up to 40 times its own weight. In addition to its wing physiology, the butterfly is known for behaviors such as long migrations and lekking. Greta oto closely resembles Greta andromica.
Web Search Results
- Project Glasswing - Anthropic
## What is Project Glasswing Project Glasswing is an initiative to secure the world’s most critical software for the AI era. We're partnering with the organizations responsible for the infrastructure billions of people depend on, and giving their defenders a head start with our newest frontier model, Claude Mythos Preview. The initiative brings together Amazon Web Services, Anthropic, Apple, Broadcom, Cisco, CrowdStrike, Google, JPMorganChase, the Linux Foundation, Microsoft, NVIDIA, and Palo Alto Networks as launch partners. [...] Image 4: CISO at Amazon Web Services logo [...] to modernize cybersecurity stacks everywhere. We commend Anthropic for partnering with the industry to ensure these powerful capabilities prioritize defense first.
- Project Glasswing: The 10 Consequences Nobody's ...
when governance gaps surface under customer or regulatory pressure. [...] to MDR — with an emphasis on the “response” part of MDR — to come to other security domains. [...] ### 3. Anthropic is now the most important partner for every security company
- What is Project Glasswing and Why Should We Care?
view as opposed to over-regulation or outright bans proposed by some ill-informed UN agencies. [...] delegations in Geneva and negotiation training for new Human Rights Council members help level the playing field for governments too. [...] #### Health Programming to Save Lives
- Project Glasswing: Securing critical software for the AI era - Anthropic
to modernize cybersecurity stacks everywhere. We commend Anthropic for partnering with the industry to ensure these powerful capabilities prioritize defense first.” [...] Mythos Preview found a 27-year-old vulnerability in OpenBSD—which has a reputation as one of the most security-hardened operating systems in the world and is used to run firewalls and other critical infrastructure. The vulnerability allowed an attacker to remotely crash any machine running the operating system just by connecting to it; It also discovered a 16-year-old vulnerability in FFmpeg—which is used by innumerable pieces of software to encode and decode video—in a line of code that automated testing tools had hit five million times without ever catching the problem; [...] We are hopeful that Project Glasswing can seed a larger effort across industry and the public sector, with all parties helping to address the biggest questions around the impact of powerful models on security. We invite other AI industry members to join us in helping to set the standards for the industry. In the medium term, an independent, third-party body—one that can bring together private- and public-sector organizations—might be the ideal home for continued work on these large-scale cybersecurity projects. ## APPENDIX
- Introducing Project Glasswing: Giving Maintainers Advanced AI to Secure the World's Code
Open source security has historically been a thankless task. Triaging and fixing bugs, writing and testing patches, crafting careful communications strategies – none of this is what maintainers had in mind when they sent their first project commit. We believe AI can help address this. When it works, AI provides scale and leverage. The new generation of frontier models can analyze code at a broad scale and at high velocity and detect patterns based on previous bug fixes. [...] Equally important, early indications point to Claude Mythos Preview and other advanced AI models not only finding vulnerabilities but also providing viable patches. When I recently spoke with the Linux Project’s Greg Kroah-Hartman, he was initially skeptical, but more recently, he has told me that some of the patches generated by AI tools were “pretty good” – which is high praise, coming from him. [...] The combination of the ability to identify and patch, on a broad scale and at a faster pace, can dramatically reduce the security burden on maintainers. Project Glasswing will allow maintainers alongside critical enterprises to improve their project security and to spend more time focusing on advancing their project code instead of playing defense. For all of us, it’s a win-win: safer code, faster software development and more incentive to become a maintainer. Making powerful AI accessible to maintainers of critical software