New Delhi: Just over a day after Anthropic hailed India as its “second-largest market” and unveiled a partnership with Tata Consultancy Services (TCS), Indian developers, enterprises and researchers found themselves locked out of the company’s most powerful artificial-intelligence tier — Claude Fable 5 and Claude Mythos 5 — following a US govt export-control order.
The two use the same underlying model and are distinguished only by their safeguards: Fable 5 is the safeguarded version released to the public, while Mythos 5, with the restrictions lifted, had been reserved for a small group of vetted partners.
Cybersecurity experts warn that the the US govt move could undermine defensive security efforts, even as they accelerate calls for sovereign AI capabilities.
In a statement on Saturday, Anthropic said the US govt, citing national security authorities, had ordered the company to suspend access to Fable 5 and Mythos 5 for all foreign nationals, both inside and outside the US. The company said it was working to restore access and that it believed the move stemmed from a misunderstanding.
The decision has quickly become a flashpoint in the broader debate over whether frontier AI models should be treated as strategic technologies subject to export controls, similar to advanced semiconductors and other dual-use technologies.
Jaspreet Bindra, co-founder and CEO of AI & Beyond, said, “In cybersecurity, the same model can be both a weapon and a shield.” While restrictions could reduce offensive cyber risks, particularly from less sophisticated actors, he cautioned that a blanket ban could also weaken legitimate defenders. “The question is who gets access, under what controls and with what auditability.”
Neehar Pathare, managing director, and chief information officer of 63SATS Cybertech, was more critical, describing the move as a “geopolitical seizure” rather than a security safeguard. Pathare argued that cybercriminals are unlikely to depend on commercial AI services like Claude and can instead turn to open-source or locally hosted alternatives.
The issue comes at a time when AI models are increasingly being used to automate vulnerability discovery, software testing and security analysis. Pathare cited Anthropic’s earlier claims that advanced cyber-capable models could accelerate vulnerability research, helping security teams identify weaknesses before they are exploited.
The development echoes a warning issued just weeks ago in JPMorganChase Center for Geopolitics’s report, Beyond the Benchmarks: A Systemic View of US-China AI Competition, which argued that AI has become “the whole ballgame” geopolitically and identified Anthropic’s Mythos as a model whose significance extends beyond technology into national security. On Friday, that reading turned to reality.
For India, which is rapidly positioning itself as a major AI market and innovation hub, the episode raises difficult questions. The country has the talent, the demand and increasingly the investment. Yet the sudden loss of access to Anthropic’s most advanced models shows participation in the AI revolution may also be shaped by strategic competition.
