Singapore and India Banks Race to Contain Anthropic's Mythos AI Cyber Threat

Banks Sound the Alarm

Financial regulators and banking associations across Asia and Europe are escalating their cybersecurity responses to Anthropic's frontier AI model, Claude Mythos Preview, amid growing concerns the technology could be weaponised against critical financial infrastructure. The Association of Banks in Singapore confirmed it is working with member banks to monitor emerging threats, share intelligence, and bolster incident response capabilities.

Singapore Coordinates a National Response

The Monetary Authority of Singapore has been coordinating with the Cyber Security Agency of Singapore since mid-April to harden defences at critical infrastructure operators, including lenders. A MAS spokesperson urged financial institutions to redouble efforts to strengthen security, proactively identify and close vulnerabilities, and raise vigilance on cyber hygiene, including timely security patching.

India Convenes Emergency Meeting

In India, Finance Minister Nirmala Sitharaman convened a high-level meeting on 23 April with SBI Chairman C.S. Setty, the IT Ministry, the Reserve Bank of India, NPCI, and CERT-In to assess Mythos-related risks. A panel under Setty has been formed to recommend mitigations, while the Indian Banks' Association has been directed to establish a real-time threat intelligence sharing mechanism with CERT-In. India's cybersecurity agency issued a high-severity advisory warning that emerging AI tools can carry out complex cyberattacks with almost zero human effort.

A Model Too Dangerous to Release

Anthropic unveiled Claude Mythos Preview on 7 April under Project Glasswing, a restricted programme granting 12 partner organisations access for defensive research. The model can autonomously identify previously unknown software vulnerabilities, generate working exploits, and execute complex cyber operations with minimal human input. During testing, Anthropic claims Mythos uncovered thousands of zero-day vulnerabilities, many critical and some one to two decades old. Independent testing by the UK AI Security Institute scored the model at 73 per cent on expert-level hacking tasks. Anthropic withheld a broader rollout, calling Mythos far ahead of any other AI model in cyber capabilities, and is now planning to extend access to European banks.

Europe Joins the Response

ESMA Chair Verena Ross warned that cyberattacks are growing in both frequency and speed as AI heightens risks for the financial sector. The Bank of England's co-chaired financial sector group said the UK is prepared for threats from Mythos and similar models, reflecting what the World Economic Forum called a shift in frontier AI from productivity to critical infrastructure and financial stability.