Finance Minister Nirmala Sitharaman met with top banking leaders, the RBI, and MeitY officials on Thursday to address urgent cybersecurity concerns. The meeting focused on the potential threat posed by Anthropic’s “Mythos” model to the integrity of the Indian financial system.
The Catalyst: Anthropic’s Claude Mythos
The discussions were triggered by reports regarding Mythos, a highly advanced AI model developed under Anthropic’s “Project Glasswing.” Key concerns include:
-
Superior Cyber Capabilities: Mythos has demonstrated the ability to outperform humans in identifying and exploiting vulnerabilities, including decades-old bugs in major operating systems and web browsers.
-
Unauthorized Access: Reports indicate that unauthorized access was made to this model, which Anthropic had previously deemed “too dangerous” for general public release.
-
Data Vulnerability: Regulators are concerned that such a powerful tool could be misused to breach secure financial databases or compromise customer funds.
Government and Regulatory Response
The Finance Ministry and the RBI are currently assessing the specific “extent of risk” facing the Indian banking sector. During the meeting, the following directives were issued:
-
Preemptive Safeguards: Banks were urged to immediately harden their digital infrastructure to protect customer data and financial assets.
-
Defensive Cybersecurity: Discussion revolved around utilizing unreleased AI previews—like the Claude Mythos Preview—strictly for defensive purposes under controlled initiatives.
-
Cross-Agency Collaboration: The Ministry is working closely with the Ministry of Electronics and Information Technology (MeitY) to monitor the global situation regarding AI-driven security breaches.
Context: Balancing Innovation with Security
While the government continues to push for AI integration in banking to improve efficiency, this meeting highlights a growing “conviction gap” regarding the safety of frontier models.
“Banks have been urged to take preemptive measures to secure their systems, data, and the money of customers.” — Ministry Official
The meeting underscores a shift toward proactive AI governance, where the focus is not just on the benefits of automation, but on the existential risks that advanced “agentic” AI models may pose to national financial stability.
