Anthropic has significantly expanded its cybersecurity initiative, Project Glasswing, by granting access to its powerful AI model, Claude Mythos, to approximately 150 new organizations worldwide.
This brings the total number of participants in the program to around 200, spread across more than 15 countries.
Focus on Critical Infrastructure
The expansion specifically targets industries vital to national and global security. Organizations in power, water, healthcare, communications, and hardware sectors are the primary beneficiaries. A cyberattack on these infrastructures could potentially affect up to 100 million people, making their defense a top priority. Each organization is required to meet strict security standards before gaining access to the model.
The Power of Claude Mythos
Claude Mythos is a frontier model equipped with advanced coding and vulnerability-hunting capabilities. Initially restricted to a small group of about 50 organizations in April 2026 to prevent potential misuse, the model has already proven its worth. Since Project Glasswing's inception, partners have reportedly used Mythos to identify more than 10,000 high- or critical-severity security flaws in major software systems.
Market Signal
The cautious yet substantial rollout of Claude Mythos to 150 critical infrastructure organizations signifies a major shift towards integrating advanced AI vulnerability analysis into the core of global defense strategies. While the model remains in a restricted "Preview" state for these partners, Anthropic expects to release "Mythos-class" models to all customers in the coming weeks, provided robust safeguards are in place.
Source: TechCrunch & Industry Reports