Anthropic Takes Project Glasswing Global to Test Claude Mythos
In Focus
- Anthropic has added 150 partners to Project Glasswing
- The new partners are drawn from under-represented sectors like healthcare and energy
- Initial Glasswing partners uncovered thousands of software vulnerabilities with Mythos
Anthropic is adding 150 partners across 15 countries to Project Glasswing to test the Claude Mythos Preview AI model. The new partners are drawn from sectors that were not represented adequately in the initial batch. These include water, power, communications, healthcare, and hardware.
“This expansion is the next step toward our long-term goals: for AI to make all software more secure, and for us to help the industry adjust to how AI could change many of the core assumptions of cybersecurity,” Anthropic said in a statement posted on its website.
What is Anthropic’s Project Glasswing?
Project Glasswing is Anthropic’s cybersecurity AI program under which the AI firm collaborates with partners to secure mission critical software. Anthropic, which focuses on enterprise customers, introduced the Glasswing in early April, 2026 and allowed 50 partners initial access to Claude Mythos Preview.
Tech giants Google, Cloudflare, Mozilla, Apple, Microsoft, Amazon Web Services, and Nvidia were among the first partners to join Anthropic’s Project Glasswing. Last month, Anthropic reported that most Glasswing partners had found “hundreds or critical-or high severity vulnerabilities in their software.”
In total, the company said its partners found over 10,000 security flaws. Several partners reported a significant increase in the speed with which they identify bugs. Following its release, Mythos sparked security concerns last month over fears that hackers could use the powerful AI model to expose software vulnerabilities. These concerns led to discussions between the U.S. government, financial institutions, and tech companies.
New Partners Must Meet Security Requirements
Anthropic said that the new partners will have to meet security requirements before accessing Claude Mythos under Project Glasswing. According to the AI firm, a majority of the new partners maintain codebases that are used by other organizations and governments.
“What each partner has in common is that a successful attack on their codebase could be catastrophic. For most partners, we estimate that a major attack could affect more than 100 million people, with important ramifications for both global and national security,” the company added.
Anthropic expanded its Mythos AI partners a day after it agreed to give the European Union access to the cybersecurity model. Although the terms and conditions under which the EU will access Mythos have not been clarified, access will be granted to the bloc’s cybersecurity agency, the European Union Agency for Cybersecurity (ENISA).
Anthropic’s IPO Progress
Anthropic expanded its cybersecurity program a day after it filed its Initial Public Offer prospectus with the Securities Exchange Commission. The move places it ahead of OpenAI, which is also preparing to list publicly this year. Recently, Anthropic became the world’s most valuable AI company following a successful $65 billion Series H funding round that pushed its valuation to close to $1 trillion. Its IPO filing paves the way for a significant share sale
