A US directive forced Anthropic, a leading US AI lab, to lock its newest large language models, Fable 5 and Mythos 5, behind a US only wall, and Anthropic had publicly named India a top two global market.
On Friday June 12, a U.S. government directive forced Anthropic, one of America's leading AI labs, to suspend access to its newest flagship models, Fable 5 and Mythos 5, for every foreign national in every country, including the company's own overseas employees. The order landed four days after Anthropic had publicly named India as a top-two global market for those same models. For the Indian founders, investors, and enterprise buyers who had been building roadmaps on leading American AI systems, the weekend question became sharper than the abstract sovereignty debate they had been having for months: what does an entire country's AI plan look like the morning after Washington pulls the access rug?
TechCrunch, citing unnamed "some reports," attributes the original security concerns about the new models to Amazon CEO Andy Jassy, a detail that has not been independently corroborated. Anthropic has publicly disputed the government's overall characterization of the order, and the White House has reportedly taken the position that restrictions of this kind are unlikely to widen to other U.S. AI labs, per TechCrunch's reporting on the suspension. What remains unstated is the directive's scope, statutory basis, duration, and issuing agency, which is why Anthropic's claim that the move is narrowly targeted is doing so much work for any Indian customer that had been assuming access to top-tier American AI systems was a stable assumption.
The market context is why the directive lands here harder than almost anywhere else. Anthropic and its Indian enterprise partner Tata Consultancy Services had recently announced a partnership to expand adoption of Anthropic's models across TCS's enterprise customer base, and India sits among the top three global AI markets by developer base and adoption, a size both companies had publicly cited. A short suspension anywhere would be an inconvenience. A suspension that lands inside the launch window of a major enterprise rollout is a stress test for the assumption that India can build on American AI systems without an American policy veto.
The reaction on the ground is a recalculation, not a rupture. Indian founders who had been picking Fable 5 or Mythos 5 for production features are now asking their engineering leads what an open-source fallback would cost in latency and quality, and whether the procurement team needs to add a clause requiring a model that runs on Indian-controlled infrastructure to the next vendor review. Investors are stress-testing their AI infrastructure theses against a scenario where access to top-tier American AI is a U.S. foreign-policy variable. And in New Delhi, policy staff who had been drafting a softer, productivity-focused AI framework are now being asked to produce a sovereignty response that does not crater enterprise productivity. None of those decisions were on anyone's calendar a week ago. All of them are now.