Anthropic expands Claude Mythos access worldwide, adds India and 14 other countries

Anthropic has expanded access to its cybersecurity-focused AI model, Mythos, to around 150 additional organisations across more than 15 countries. The move comes under the company’s Project Glasswing initiative, which was launched earlier this year to help organisations identify and fix software vulnerabilities before they can be exploited by attackers. Also Read: Florida sues OpenAI and Sam Altman: State claims ChatGPT is unsafe for children

According to the company, the latest expansion takes the total number of participating organisations to roughly 200. Reports suggest countries gaining access include India, Australia, Belgium, Canada, France, Germany, Italy, Japan, New Zealand, the Netherlands, Spain, Switzerland, Sweden, and South Korea, among others. Also Read: Beyond ChatGPT: OpenAI begins hiring to build personal robots

What is Anthropic Mythos?

Mythos is Anthropic’s specialised AI model designed for cybersecurity tasks. Unlike general-purpose AI chatbots, the model focuses on finding vulnerabilities in software code, including security flaws that may otherwise go unnoticed. Also Read: Anthropic launches Claude Opus 4.8 with focus on honesty, better judgement, and lower AI costs

The company claims Mythos can identify large numbers of vulnerabilities across complex codebases over a relatively short period. It forms the foundation of Project Glasswing, which brings together governments, technology firms, infrastructure providers, and security organisations to improve software security.

Anthropic introduced Project Glasswing in April 2026 with an initial group of 50 partners. That first batch included major organisations such as Amazon Web Services, Apple, Google, Microsoft, NVIDIA, Cisco, CrowdStrike and JPMorgan Chase.

New industries now included

One of the biggest changes in this expansion is the addition of sectors that were not heavily represented in the initial rollout.

The new group includes organisations working in power, water, healthcare, communications, and hardware. According to Anthropic, many of these organisations maintain software and codebases that are relied upon by governments, businesses, and millions of users worldwide.

The company said that a successful cyberattack on some of these systems could affect more than 100 million people, making them important from both national security and public infrastructure perspectives.

Reports have also suggested that organisations such as Okta, Samsung, SK Telecom, NATO, and ENISA are among those receiving access, although Anthropic has not publicly confirmed the full participant list.

Focus on software vulnerabilities

Anthropic says organisations already participating in Project Glasswing have identified more than 10,000 vulnerabilities that were classified as high or critical severity.

The company is also exploring ways to speed up the review and patching process for open-source software projects. In addition, it is working on best practices for reporting vulnerabilities to developers so that issues can be fixed more efficiently.

Add Techlusive as a Preferred Source

The expansion comes at a time when AI companies are increasingly building cybersecurity-focused models. Anthropic recently said it expects other companies to develop similar capabilities soon. Rival OpenAI has already introduced its own cybersecurity-focused AI model, GPT-5.5-Cyber, for partner testing.