Anthropic halts new models India weighs AI future

Anthropic’s decision to suspend access to its latest AI models has sparked fresh conversations about India’s reliance on foreign technology. The U.S. government’s directive, which restricts use of the Fable 5 and Mythos 5 models for non-U.S. citizens, came just days after the company partnered with Tata Consultancy Services to boost AI adoption in India.

The move demonstrates how closely India’s ambitions in artificial intelligence are tied to systems developed abroad. The directive followed reports of security concerns raised by Amazon CEO Andy Jassy, according to the outlet. While the White House has not extended similar restrictions to other firms, it has privately criticized Anthropic’s handling of vulnerabilities in its models. The company has rejected the government’s claims, arguing the action was unwarranted.

India’s tech community is now reevaluating its long-term strategy. Some argue the episode highlights the risks of depending on a few foreign AI providers, while others warn that geopolitical decisions can abruptly limit access to critical tools. The country, already a key market for companies like Anthropic and OpenAI, faces a growing dilemma: how to balance innovation with self-reliance.

“This changes everything,” said Aakrit Vaish, founder of Activate, a platform supporting Indian AI startups. He called the move a wake-up call for the sector, predicting a shift toward open-source models and reduced dependence on U.S. systems. Startups like Atomicwork, which operates across the U.S. and India, worry about competitive disadvantages if access to advanced AI tools becomes uneven.

Vijay Rayapati, CEO of Atomicwork, noted that teams with non-U.S. members may struggle under new restrictions. The company’s engineering team is based in Bengaluru, but its U.S. presence complicates compliance. Meanwhile, U.S. firms like Opendoor have scaled back operations in India, citing shifts in AI strategy that could reshape how global talent is distributed.

Indian tech leaders are calling for more investment in domestic AI capabilities. Zoho’s founder, Sridhar Vembu, urged companies to adopt smaller, open-source models, including those from China. Mohandas Pai, a former Infosys executive, proposed a ₹500 billion annual fund for AI and a ₹2 trillion credit program to support infrastructure. His plan dwarfs India’s current efforts, which include a $1.2 billion initiative to boost AI development.

Yet challenges remain. Hemant Mohapatra of Lightspeed argues that talent, computing resources, and execution—rather than funding—are the biggest hurdles. Training a frontier AI model can cost hundreds of millions to billions, he noted, but successful firms often scale as adoption grows.

For policy experts, the incident may reinforce concerns about strategic autonomy. Prasanto Roy, a Delhi-based analyst, compared the situation to Russia’s loss of SWIFT access after the Ukraine invasion. He warned that American AI models are inherently linked to U.S. geopolitics, and the episode could fuel nationalist sentiment in India.

India’s AI ecosystem remains focused on applications built atop existing models, rather than foundational research. Startups like Avataar AI are developing specialized tools, but few are pursuing large-scale AI development. Krutrim, once a key player in foundational models, has shifted to infrastructure services instead.

The answer may determine India’s path toward a self-sufficient AI industry, influencing its tech sector, economic strategies, and global positioning.