For the past two years, artificial intelligence companies have competed to build smarter chatbots. But behind the scenes, a more consequential race is unfolding.
The battle is now over cybersecurity.
Anthropic, the company behind Claude, is expanding access to its highly restricted cybersecurity model, Claude Mythos Preview, to organisations across more than 15 countries, including India. The move brings one of the world’s most closely guarded AI systems to a country that sits at the heart of the global software industry.
That matters because the next chapter of AI may not be about generating text or images. It may be about defending the digital infrastructure that powers modern economies.
What is Anthropic’s Mythos
Claude Mythos Preview is not a conventional chatbot.
Anthropic introduced the model in April 2026 under Project Glasswing, a programme designed to help secure critical software systems.
According to Anthropic, the model has already discovered thousands of previously unknown vulnerabilities across major operating systems, browsers and widely used software components. The company cited examples that underline the model’s capabilities.
Mythos uncovered a 27-year-old vulnerability in OpenBSD that allowed attackers to remotely crash systems. It also identified a flaw in FFmpeg, a widely used multimedia framework, that had gone unnoticed despite automated testing tools executing the affected code millions of times.
Anthropic also said the model autonomously chained together vulnerabilities in the Linux kernel to escalate privileges on a machine.
Because of these capabilities, Anthropic initially limited access to a small group of trusted partners rather than releasing the model broadly.
Project Glasswing in India
That circle is now getting larger.
Reuters reported that Anthropic plans to expand Project Glasswing from roughly 50 organisations to around 200 organisations spread across more than 15 countries. About 150 new participants are being added in the latest phase.
Business Standard reported that India is among the countries gaining access as the programme broadens beyond its largely US-focused launch.
The expansion includes organisations working in sectors such as healthcare, communications, energy and hardware. According to reports, participants have already identified more than 10,000 severe software vulnerabilities using the system.
Anthropic has argued that advanced AI can provide defenders with a lasting advantage against increasingly sophisticated cyber threats.
Why India Matters
India’s inclusion is significant because of the country’s central role in the global technology ecosystem.
Indian companies manage and support digital infrastructure used by enterprises, banks, telecom operators and governments around the world. The country is also home to one of the world’s largest pools of software engineers and cybersecurity professionals.
Access to Project Glasswing does not mean the technology becomes publicly available in India. Instead, selected organisations will be able to use the system under Anthropic’s controlled framework.
That distinction is important.
Mythos remains one of the few frontier AI models whose deployment has been tightly restricted because of concerns that the same capabilities that help defenders identify vulnerabilities could also be misused if safeguards are absent.
Global Security Stakes Rise
Anthropic’s expansion comes as governments and companies are increasingly treating AI-enabled cybersecurity as a strategic issue.
Concerns surrounding Mythos initially centred on fears that such systems could dramatically accelerate offensive hacking. However, those fears have appeared more limited as the model has been used under controlled conditions.
The European Commission recently confirmed that it had held discussions with Anthropic over possible future access for European cybersecurity bodies.
South Korea’s Science Ministry also announced that the Korea Internet and Security Agency had joined Project Glasswing alongside major domestic companies.
These developments suggest that governments and institutions increasingly see advanced AI systems not simply as productivity tools but as part of their cybersecurity infrastructure.
Beyond Chatbots And Search
The significance of India’s inclusion extends beyond Anthropic. It signals that the AI race itself is changing.
For years, success in artificial intelligence was measured by who built the best chatbot. Increasingly, it may be measured by who can secure critical software, protect infrastructure and respond to emerging digital threats.
Anthropic’s decision to extend Project Glasswing to India reflects that shift. The future of AI will not be determined solely by what these systems can create. It will also depend on what they can protect.
Also Read: Why Global Investors Are Racing To Build AI Data Centres Around India’s Financial Capital
