RBI reaches out to global regulators for risk assessment on Anthropic Claude Mythos: Report

Amid the Claude Mythos storm, the Reserve Bank of India (RBI) is reportedly engaging with international regulators, domestic lenders, and government officials to evaluate the potential cybersecurity threats posed by Anthropic's newly released AI model Mythos.

According to a report from Reuters, the Reserve Bank of India (RBI) has conducted a preliminary assessment that aligns with concerns raised by regulators worldwide. Officials believe the advanced AI system could significantly accelerate the discovery and exploitation of software vulnerabilities, potentially increasing the risks to India's financial sector.

RBI coordinates with global banks, regulators

In the report, it is said that RBI officials have held consultations on Mythos-related risks with counterparts at the US Federal Reserve and the Bank of England, among others. The central bank may also seek direct engagement with Anthropic, the sources said.

"Globally, we are discussing with other countries and other regulators on what are the developments and what safeguards need to be taken," one source told Reuters.

Regulators across Asia, Europe, and the United States have already warned banks to review their defences and preparedness against potential threats from the AI model. In Japan, the financial watchdog is scheduled to meet with banks this week, while central banks in Australia and New Zealand are closely monitoring developments.

NPCI seeks early access to Claude Mythos

On the sidelines, India's National Payments Corporation of India (NPCI) is working alongside a small group of banks to secure early access to Mythos. The goal is to proactively identify vulnerabilities and "day-zero" cyber risks before any wider deployment from Anthropic. The NPCI currently nurtures the Unified Payments Interface (UPI) technology, which has been hailed as one of the most secure digital payment platforms available today.

However, obtaining access to Mythos could be challenging, since Mythos is hosted on strictly controlled servers in the United States. Conducting tests on Mythos involving Indian customer data in foreign jurisdictions raises compliance issues, as another source raises concern.

Access to Claude Mythos has so far been tightly restricted. Anthropic had initially limited the availability of Mythos to a small number of US organisations responsible for critical digital infrastructure. It is said that Anthropic plans to extend access to European banks in the near future, but nothing is concrete at the moment.

RBI preparing long-term AI guidelines

The RBI is also working on broader guidelines for banks entering into enterprise partnerships with advanced AI models, including Mythos and Anthropic's Claude family. These guidelines form part of a longer-term strategy to ensure a safe AI adoption in India's financial system. The discussions are still said to be in early stages, but the central bank is expected to enforce strict compliance with its 2018 data localisation rules.

For those unaware, these regulations currently require payment system providers to store all end-to-end transaction data -- including user information and payment messages -- exclusively on servers located within India.