Anthropic has announced a massive global expansion of its “Project Glasswing” initiative, extending access to its highly advanced Mythos AI model to 150 organizations across more than 15 countries, including India.
Mythos is a specialized AI engine built to detect and patch complex software vulnerabilities. Because of its raw power and advanced coding capabilities, Anthropic had previously kept the model heavily restricted out of fear that it could be weaponized for malicious hacking. The global rollout aims to fortify defenses before the software can be exploited by bad actors.
Global Rollout: Who is Getting Access?
The expansion broadens the program’s reach significantly beyond its initial, heavily restricted pool in the United States and the United Kingdom.
Newly Admitted Countries & Alliances:
-
Asia-Pacific: India, South Korea, and Japan.
-
The Five Eyes Alliance: Canada, Australia, and New Zealand (joining the US and UK).
-
Europe: France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, and Sweden.
-
International Bodies: The North Atlantic Treaty Organisation (NATO) and the European Union Agency for Cybersecurity (ENISA).
Critical Infrastructure Operators in Focus
Anthropic’s expansion targets global companies where a system failure would trigger a massive domino effect. According to the company, a successful cyber attack on any of these partners’ software codebases could affect more than 100 million people, risking catastrophic national and economic security failures.
The 150 selected organizations now span previously underrepresented sectors like power, water utilities, healthcare, hardware manufacturing, and telecommunications. Notable new corporate participants include:
-
Tech & Hardware: Samsung Electronics, SK Hynix, SK Telecom, and identity management platform Okta.
-
Financial Infrastructure: International payments giant SWIFT, Euroclear, and the Intercontinental Exchange.
The Elite Club: These new members join early-phase tech and financial giants who already had access, including Microsoft, Apple, Oracle, JPMorgan Chase, and CrowdStrike.
The Strategic Shift Behind Mythos
When Anthropic launched Claude Mythos Preview in April, it initially limited access to a tight circle of roughly 50 mostly US-based entities. This strict gating sparked anxiety among international banks, regulators, and foreign governments who worried they were being left vulnerable to security threats Anthropic was uncovering.
| Phase | Timeline | Organization Count | Primary Geographic Focus |
| Initial Launch | April | ~50 Organizations | Primarily United States & United Kingdom |
| Expansion Phase | June (Current) | 150 Organizations | 15+ Countries (India, Eurozone, Five Eyes, NATO) |
Looking ahead, Anthropic plans to scale Project Glasswing even further. Future rounds will prioritize critical open-source software maintainers and global safety-testing organizations.
Note: This major security rollout comes on the heels of another massive milestone for the AI firm, following Anthropic’s recent filing for an Initial Public Offering (IPO) that could see the company valued at over $1 trillion.
