India Eyes Advanced AI Access with US & Anthropic

The Indian Centre is reportedly in negotiations with the United States and Anthropic, a prominent AI company, to devise a system that would allow Indian businesses access to Anthropic’s sophisticated AI model, Claude Mythos. This initiative is part of a broader strategy to bolster the security of India’s vital digital infrastructure, including power grids, telecommunications, and banking systems.

According to a senior official cited by ET, the government is actively exploring various “mechanisms” and coordinating the “logistics” to ensure that Indian companies can obtain equitable access to the AI model without jeopardizing the nation’s critical infrastructure. The official noted the urgency, stating, “The government needs to build its capacity as of yesterday,” highlighting the potential for rapid, unannounced releases of such powerful models by other companies.

Earlier, Finance Minister Nirmala Sitharaman had voiced concerns regarding the risks associated with the model, indicating that the Ministry of Electronics and Information Technology (MeitY) was actively engaged with the US administration, Anthropic, and system testers. Sitharaman reportedly remarked on the significant cybersecurity challenges that Mythos could present.

In response to these concerns, the Finance Minister, alongside IT Minister Ashwini Vaishnaw, recently chaired a meeting with banking institutions and key cyber agencies to assess the potential impact of Mythos on India’s financial and digital landscape. The meeting was attended by senior officials from various financial bodies, including the Reserve Bank of India (RBI), MeitY, the National Payments Corporation of India (NPCI), and the Department of Financial Services.

During the discussions, the Finance Minister urged the Indian Banks’ Association (IBA) to develop a coordinated response mechanism for threats related to the model and advised banks to enhance their cybersecurity measures by collaborating with specialized agencies and experts. Banks were also instructed to promptly report any suspicious cyber incidents to agencies like CERT-In. Furthermore, CERT-In, the National Critical Information Infrastructure Protection Centre (NCIIPC), and the financial sector were directed to reinforce the security of critical infrastructure, such as power grids, telecom networks, and banking systems.

The RBI has also been engaged in separate assessments of the risks posed by Mythos, collaborating with global regulators, Indian banks, and government officials. Meanwhile, Anthropic is understood to be holding separate discussions with Indian authorities at its US headquarters.

Last week, Nasscom, a technology industry association, formally requested access for Indian technology companies under ‘Project Glasswing’. Nasscom emphasized that Indian firms manage critical code utilized by organizations globally and require access to Mythos to bolster cybersecurity resilience.

Launched in April, Claude Mythos is specifically designed for advanced coding and cybersecurity applications. It possesses the capability to identify and exploit critical software vulnerabilities across major operating systems and browsers, including those that may have remained undetected for extended periods. Due to its potential for misuse, Anthropic has restricted access to the model under Project Glasswing, and no Indian company has been included to date.

Anthropic’s latest AI model has raised concerns among governments worldwide due to its ability to uncover and exploit vulnerabilities that could potentially compromise power grids, telecom networks, banking systems, and defence infrastructure, exposing them to cyberattacks. Consequently, several nations are implementing stricter controls on access to the model, viewing it as both a valuable cybersecurity tool and a potential national security risk.