The U.S. decision to restrict access to Anthropic’s Mythos and Fable large language models has prompted national security officials in India to highlight the need for domestic AI development. Officials have faced limited support while advancing sovereign AI initiatives.
Sridhar Vembu, founder of Zoho and a member of India’s National Security Advisory Board, stated on social media that globalization has ended and India must pursue independent progress. He suggested that organizations adopt smaller open-source models from India and China.
The U.S. government directed Anthropic to disable access to the models for non-U.S. users, including within the company. Mythos is described as effective at identifying cybersecurity vulnerabilities missed by human researchers.
India had sought access to Mythos through Project Glasswing amid concerns over AI-related cyber threats. The new restrictions may affect that access. The Ministry of Electronics and Information Technology and the Ministry of External Affairs did not respond to inquiries. The Indian Cybercrime Coordination Centre and CERT-in also did not comment.
Other projects using Fable have been disrupted. Entrepreneur Vikram Chandra noted that ongoing work relying on the model would stop.
India trails China and the U.S. in frontier AI training capacity. While Chinese firms use older GPUs at scale, India faces constraints in chips, data centers, and power. Building a domestic alternative would require substantial investment, which remains limited.
Vembu called for deeper research but noted high costs and GPU export limits. He supported lower-cost efforts, including Zoho’s recent server development. T.V. Mohandas Pai advocated for larger government funding, including an annual fund for deep tech and AI.
