Anthropic access ban sparks India debate over sovereign AI ambitions

Anthropic’s suspension of Claude Fable 5 and Claude Mythos 5 for any foreign national has turned a customer-access dispute into a broader test of India’s AI sovereignty. The models had launched only on June 9, and the cutoff came after a U.S. export-control directive that applied to foreign nationals inside or outside the United States.

Anthropic said it was “working to restore access as soon as possible” and later said it disagreed with the directive, but the disruption landed at a sensitive moment. The company had described Fable 5 as a Mythos-class model with stronger cybersecurity safeguards, while Mythos 5 was limited to a small group of vetted customers through trusted-access programs. Anthropic has also argued publicly for tighter U.S. export controls on advanced AI chips and model weights, underscoring how frontier AI is increasingly treated as a national-security asset.

In India, the reaction quickly moved beyond one company’s product policy. Zoho co-founder and former chief executive Sridhar Vembu called the development a wake-up call and said “globalization is dead,” urging India to build its own AI future and rely less on foreign frontier models. Investor Mohandas Pai has also pushed for a stronger India AI effort. The common concern is straightforward: if access to cutting-edge models can be altered by a foreign government order, startups, researchers and policymakers in India remain exposed to decisions made far beyond New Delhi.

That vulnerability is why the IndiaAI Mission has become central to the sovereignty debate. The Union Cabinet approved the mission on March 7, 2024, with a five-year outlay of Rs 10,371.92 crore. The plan includes public AI compute infrastructure of 10,000 or more GPUs and support for indigenous foundational models, while the Union Budget 2024-25 separately set aside Rs 551.75 crore for the mission. Later government backgrounders said India had deployed 38,000 GPUs and framed affordable, trusted AI infrastructure as a national priority.

The government has cast the effort as part of a larger push to make AI in India and make AI work for India. The Anthropic episode now serves as a stress test for whether that ambition is moving from policy language to industrial capacity. For India, the real question is whether foreign model restrictions will trigger sustained investment in domestic compute, models and technical depth, or simply add another slogan to a familiar debate about self-reliance.