Anthropic Expands Mythos Access to 150 Organizations

According to Anthropic's blog post, the company is expanding Project Glasswing by granting access to approximately 150 additional organizations for Claude Mythos Preview, its model for detecting software vulnerabilities. Anthropic says the initial cohort of roughly 50 partners has already uncovered more than 10,000 high- or critical-severity security flaws. Bloomberg reports the expansion brings the total number of groups with access to about 200. Per Anthropic and CNBC, the new partners are based in more than 15 countries and span sectors such as power, water, healthcare, communications, and hardware; Anthropic says each organization must meet its security requirements before gaining access, and estimates that a successful attack on a typical partner's codebase could affect more than 100 million people.

Anthropic has separately reported scanning more than 1,000 open-source projects with Mythos and identifying 23,019 issues, of which 6,202 were rated high- or critical-severity. The company frames Project Glasswing as a cooperative effort with partners, open-source maintainers, and the US government, and says partners have begun using the model not only to find vulnerabilities but also to write patches and run pre-release checks.

Class B analysis: models that automate vulnerability discovery can scale detection velocity and surface issue classes that manual review misses, while also multiplying the volume of findings that require triage. Teams using advanced automated scanners typically confront false positives, reproducibility challenges, and the need to integrate results into existing bug-tracking and disclosure pipelines. Access gating, operational controls, and coordination with maintainers materially affect how quickly high-confidence issues are mitigated.

The expansion moves Claude Mythos beyond a tightly limited preview toward broader enterprise and critical-infrastructure use, amplifying both defender utility and dual-use concerns. CNBC and Bloomberg place the announcement alongside Anthropic's confidentially filed draft S-1 with the SEC, and CNBC reports Anthropic intends to offer Mythos in the European Union, underscoring commercial and regulatory dimensions to wider availability.

• •How Anthropic and partners handle disclosure timelines and coordination with maintainers and national CERTs, including whether standardized processes are published.
• •Whether partner identities are disclosed, since CNBC notes the new organizations were not named.
• •Regulatory and policy responses in the EU and national-security communities given the planned EU offering and the draft S-1.
• •Technical metrics such as reproducible-finding rates versus false positives and integration of Mythos output into remediation workflows.