India’s leading fintech companies are urging Anthropic PBC to grant them early access to Mythos, an advanced AI model that has raised global concerns over a potential surge in cyberattacks. Firms such as One97 Communications Ltd., Razorpay Software Ltd., and Pine Labs Ltd. have asked the San Francisco-based AI developer to allow them to test Mythos so they can identify and fix vulnerabilities in their own systems. This follows Anthropic’s decision to begin a restricted rollout of the model, which it believes is too risky for broad release.

Vijay Shekhar Sharma, founder and CEO of One97, said the company recently held an urgent discussion with Anthropic about when a second group of approved companies might gain access. According to him, Anthropic questioned how One97 intended to use Mythos and what value it would bring, highlighting the company’s cautious approach to distribution.

Sharma also expressed concern about the broader implications of such technology, suggesting it raises deep security fears for financial systems and national infrastructure. He warned that modern digital networks could be compromised from anywhere, making traditional forms of conflict less relevant.

The demand from Indian firms mirrors global anxiety, as regulators and financial leaders worry after reports that Mythos can uncover long-standing cybersecurity flaws. Anthropic initially tested the model internally before granting limited access to a small group of companies including Amazon Web Services, Apple, and JPMorgan Chase, and is now considering a controlled expansion under a program called Project Glasswing.

The development has unsettled global financial authorities, with US Treasury officials describing the model as a major leap in capability, while European leaders have cautioned about the risks of misuse. Experts are increasingly questioning whether such AI systems could enable large-scale financial fraud, disrupt payment infrastructure, or destabilize the global financial system. Indian regulators, including the Reserve Bank of India, have not yet commented on any formal risk assessment.

Meanwhile, companies like Razorpay are actively preparing for potential threats, with internal teams strengthening cybersecurity defenses. CEO Harshil Mathur said the company has formally requested access to Mythos to test and improve system security, describing it as a time-sensitive challenge for startups. He added that the model has become a major topic among startup communities recently, and suggested Anthropic may eventually expand access under strict usage rules focused only on security testing.

India, with its large pool of software engineers serving global financial institutions, has become a key market for Anthropic’s Claude system, widely used for coding, debugging, and IT modernization. Pine Labs CEO Amrish Rau emphasized that regulators will likely tighten cybersecurity standards in response to such threats, stressing that security must now be treated as a core priority rather than a formality.