Anthropic Expands Mythos Cybersecurity Agent to 150 New Organizations Across 15+ Countries Through Project Glasswing

Anthropic announced Tuesday that approximately 150 new organizations will gain access to Claude Mythos Preview through its Project Glasswing cybersecurity program, tripling the initiative’s reach from the roughly 50 partners that joined the initial cohort in April. The new partners span more than 15 countries and include critical infrastructure sectors that were underrepresented in the first wave, according to Anthropic’s blog post.

Scope of the Expansion

The new cohort covers power, water, healthcare, communications, and hardware, sectors Anthropic identified as gaps in the initial launch. Many of the incoming partners are vendors: companies and nonprofits that maintain codebases relied upon by other organizations, including governments. Each new partner must meet Anthropic’s security requirements before gaining access, according to the company’s announcement.

“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 wrote.

Existing major partners include Apple, Nvidia, Microsoft, CrowdStrike, and Palo Alto Networks. Cloud data management platform Rubrik disclosed it was among the new batch, according to CNBC.

Results So Far

Since April, Project Glasswing partners have surfaced more than 10,000 high or critical-severity security flaws using Mythos Preview, according to Anthropic. Partners are now using the model beyond vulnerability discovery: writing patches, running pre-release checks to prevent vulnerabilities from shipping, penetration testing, automating threat detection, and rebuilding legacy codebases in memory-safe languages.

The company estimates that a major cyberattack on any individual partner’s codebase could affect more than 100 million people, according to CNBC.

The General-Access Question

Anthropic acknowledged the core tension: Mythos-class capabilities are coming to other AI companies within 6 to 12 months, and those competitors may release them without misuse safeguards. The company said it is “working as quickly as we can to safely release Mythos-level capabilities in general access” but stated that neither Anthropic nor any other AI developer has yet built safeguards robust enough for an uncontrolled release, per the blog post.

In the interim, Anthropic plans to expand Glasswing further, prioritizing essential infrastructure providers, critical open-source software maintainers, and safety testers. The company also intends to scale its Cyber Verification Program, which grants Mythos-class capabilities for specific cyberdefense tasks.

IPO Timing

The expansion comes one day after Anthropic confidentially filed its IPO prospectus with the SEC, beating rival OpenAI to the milestone and paving the way for what could be the largest AI share sale to date, as reported by CNBC. Demonstrating that Mythos can scale across geographies and sectors without a security incident strengthens the pre-IPO narrative that Anthropic’s safety-first approach translates into defensible enterprise revenue.