What is Anthropic's Mythos AI and What Does India's Access Mean?

Claude-maker Anthropic will grant India access to its Mythos AI model, a powerful cybersecurity tool it has previously described as too risky for unrestricted public release, the Financial Times reported.

This was reported after Anthropic announced the widening accessibility to the technology on Tuesday, extending it to 150 organisations across more than 15 countries. Mythos is an AI model designed to identify and patch cybersecurity vulnerabilities. Until now, access had been largely limited to organisations in the United States and the United Kingdom.

The development also comes a day after the company filed for an initial public offering (IPO) that could value the Claude-maker at more than $1 trillion. On June 1, Anthropic announced it had confidentially submitted a draft registration statement on Form S-1 to the US Securities and Exchange Commission (SEC) for a proposed IPO of its common stock.

It did not, however, disclose the number of shares to be offered or its targeted valuation, adding that any offering would be subject to market conditions and conducted in accordance with US securities laws.

Project Glasswing Goes Global

India's inclusion is part of the broader expansion of Project Glasswing, an industry initiative that uses Mythos to detect and address security vulnerabilities.  

Other countries joining the initiative include members of the Five Eyes intelligence alliance, Canada, Australia and New Zealand, as well as France, Germany, Italy, Switzerland, the Netherlands, Spain, Belgium, Sweden, Japan and South Korea. The North Atlantic Treaty Organisation (NATO) has also been granted access.

Among the organisations receiving access are US identity and access management firm Okta, South Korean technology groups Samsung Electronics, SK Hynix and SK Telecom and major financial market infrastructure providers including Euroclear, Intercontinental Exchange and international payments network SWIFT.

The European Union Agency for Cybersecurity (ENISA) is also among the new participants, according to the report.

What Is Anthropic's Mythos AI?

Mythos is an AI model developed by the San Fransico-based startup and designed specifically to identify cybersecurity vulnerabilities in software systems. Unlike Anthropic's general-purpose Claude models, Mythos was built for a narrow but high-stakes purpose.

Anthropic announced the model's development in April but said it would not make it publicly available, citing the technology's potential to cause harm if accessed without restrictions.

Mythos was designed to analyse large volumes of code, identify weak points in IT infrastructure and flag security flaws before they can be exploited in cyberattacks.

According to Anthropic, the model can process and assess complex software environments simultaneously, enabling organisations to detect hidden vulnerabilities at a scale and speed that traditional security testing methods cannot match. The model can also assist organisations in patching vulnerabilities more quickly and strengthening their cyber defence systems.

The company has described Mythos as one of its most sensitive AI systems, capable of uncovering previously unknown software weaknesses, commonly referred to as zero-day vulnerabilities, more efficiently than conventional cybersecurity tools.

However, Anthropic has cautioned that the same capabilities could potentially be misused by malicious actors to identify weaknesses in systems, which is why access to Mythos remains tightly controlled and is extended only to vetted organisations.

Instead of a public release, Anthropic initially extended access to a limited set of technology companies in the US and the UK through Project Glasswing, which it has described as "an effort to secure the world's most critical software."

Through Project Glasswing, participating organisations use Mythos to scan critical software and infrastructure for hidden risks, helping governments and companies respond more rapidly to emerging threats.

Why Does India's Inclusion Matter?

Anthropic said the latest cohort includes operators of critical infrastructure spanning financial services, cybersecurity and technology, as well as sectors underrepresented in the initial phase of the programme, including power, water, healthcare, communications and hardware.

"What each partner has in common is that a successful attack on their codebase could be catastrophic," the company said. Anthropic estimated that, for most participating organisations, a major cyberattack could affect more than 100 million people, with significant implications for national and global security.

India is not formally part of Western intelligence alliances such as Five Eyes or NATO, both of which have also been granted access to Mythos. Its inclusion in Project Glasswing therefore marks a notable extension of the programme beyond its original geopolitical boundaries.

The decision comes at a time when India is emerging as a significant hub for AI development, digital public infrastructure and technology services, while simultaneously facing rising cybersecurity threats targeting financial institutions, government systems and critical infrastructure.

The company also said it plans to further expand Project Glasswing by prioritising additional essential infrastructure providers, maintainers of critical open-source software and safety-testing organisations worldwide.

“This expansion is the next step toward our long-term goals: for AI to make all software more secure, and for us to help the industry adjust to how AI could change many of the core assumptions of cybersecurity,” Anthropic added.

Access to an advanced AI-driven cybersecurity tool such as Mythos could help Indian organisations identify vulnerabilities and respond more quickly to sophisticated attacks.