The Indian government has initiated a comprehensive assessment of potential risks posed by advanced artificial intelligence models, such as Anthropic's Mythos. Finance Minister Nirmala Sitharaman and IT Minister Ashwini Vaishnaw chaired a high-level meeting on Thursday with banks and regulators to review emerging threats from these AI models.
High-Level Meeting on AI Threats
The meeting was attended by senior officials from the Reserve Bank of India (RBI), the National Payments Corporation of India (NPCI), the Indian Computer Emergency Response Team (CERT-In), and CEOs of various commercial banks. Officials discussed the potential misuse of such AI tools to exploit software vulnerabilities and financial systems.
Unprecedented Nature of Threats
Sitharaman described the nature of the threat from new AI models as “unprecedented” and stressed the need for greater preparedness and coordination across financial institutions. The Indian Banks’ Association was advised to create a coordinated response mechanism for AI-related threats.
Strengthening Cybersecurity
The finance minister also recommended that banks strengthen their cybersecurity systems, engage specialized agencies, and ensure real-time sharing of threat intelligence among themselves. Banks were directed to report any cyber incidents immediately to authorities such as CERT-In.
Anthropic's Mythos Model
According to Anthropic, its Claude Mythos model, currently being tested and available only to companies like Amazon Web Services, Apple, Microsoft, and Google, can independently identify and exploit software weaknesses, including previously unknown vulnerabilities in operating systems and web browsers. This raises significant concerns over potential misuse.
Opportunities and Risks
Department of Financial Services Secretary M Nagaraju stated that the technology presents both risks and opportunities. “Mythos is a threat and opportunity for the fintech ecosystem to experiment,” he said. He added that AI can improve credit access using alternative data, such as digital transactions, GST records, and telecom data, but also poses risks related to cybersecurity and data privacy.
