Anthropic said it is expanding Project Glasswing, its Claude Mythos cybersecurity program, to about 150 new organizations across more than 15 countries, most of them critical-infrastructure providers.
Anthropic said Tuesday it is expanding Project Glasswing, its controlled cybersecurity program built around Claude Mythos, to about 150 new organizations in more than 15 countries.
The company said most of the newly added participants provide critical infrastructure to large numbers of people. The expanded group includes organizations in power, water, healthcare, communications and hardware.
Project Glasswing is Anthropic's initiative for giving selected partners access to its most cyber-capable model for defensive vulnerability finding. The company described the program as a controlled effort aimed at improving software security in sensitive environments.
Anthropic first outlined Project Glasswing and Claude Mythos Preview in April. Tuesday's announcement marks a much wider rollout, and Reuters and TechCrunch both reported the expansion the same day.
The company did not disclose the full list of organizations in the new cohort. It also did not say how many total partners are now participating in Project Glasswing.
The expansion underscores how major AI developers are increasingly positioning advanced models as security tools for critical systems, while keeping access limited and targeted rather than broadly open.
Revision note
Initial automated publication.