India Reassesses AI Strategy After Anthropic Restricts Access to New Models

Anthropic’s decision to suspend access to its latest AI models following a directive from the U.S. government has sparked fresh debate across the global technology sector. In India, the development has revived longstanding concerns over whether one of the world’s fastest-growing AI markets can continue to depend on advanced technologies created and ultimately controlled outside its borders.

The announcement came late Friday after Anthropic disclosed that it had received a U.S. government directive instructing the company to suspend access to its newly introduced Fable 5 and Mythos 5 models for all foreign nationals, including foreign-national employees working within Anthropic. The timing drew particular attention because it came only days after the company unveiled a partnership with Indian IT services leader Tata Consultancy Services to accelerate enterprise AI adoption across India, highlighting how deeply the country’s AI ecosystem has become intertwined with U.S.-developed frontier technologies.

Although the full implications of the directive remain uncertain, several reports indicated that the original security concerns were first brought to the attention of U.S. authorities by Amazon CEO Andy Jassy. Separately, The Information reported that the White House is not expected to impose similar restrictions on other AI companies and has privately criticised Anthropic’s handling of alleged jailbreak vulnerabilities. Anthropic has challenged that characterisation and maintained that the government’s action was unwarranted.

Even so, the incident has sparked widespread discussion among Indian entrepreneurs, investors, and policymakers about whether the country should accelerate the development of homegrown AI capabilities, invest more aggressively in open-source models, or continue to rely on a limited number of U.S.-based frontier AI providers. For some observers, the episode serves as a warning about technological dependence. Others see it as evidence that access to increasingly important AI systems may ultimately be influenced by geopolitical considerations outside India’s control, which may ultimately influence access to increasingly important AI systems for leading AI companies. Both Anthropic and OpenAI have described India as their second-largest market after the United States, reflecting the country’s expanding influence in the global AI landscape. Over the past several months, both companies have established offices in India, increased hiring, strengthened partnerships, and launched enterprise initiatives, aiming to tap into the country’s large developer community, startup ecosystem, and business sector.

Within India’s technology industry, Anthropic’s announcement quickly became more than an isolated company decision. It reopened broader discussions about India’s long-term AI strategy and whether the country could afford to remain dependent on a small group of foreign providers supplying the world’s most advanced AI models.

“It completely changes things,” said Aakrit Vaish, founder of Indian AI venture platform Activate, referring to Anthropic’s move. “I think this materially changes the way all of us should be thinking about sovereign AI in India.”

Vaish said he woke up on Saturday “shocked and confused” by the announcement and believes the decision significantly strengthens the case for building stronger domestic AI capabilities. He expects many startups to shift toward open-source AI models and said he plans to encourage companies in his investment portfolio to reduce their reliance on a handful of frontier model providers.

For other founders, the greater concern centres on how restrictions affecting access to frontier AI models could influence global competitiveness. Vijay Rayapati, co-founder and CEO of Atomicwork, argued that the incident demonstrates the challenges faced by startups with internationally distributed teams if access to advanced AI increasingly becomes subject to geopolitical limitations.

Atomicwork employs roughly 25 people in the United States, while much of its engineering organisation operates from Bengaluru, India.

“If your AI team is not made up entirely of U.S. citizens, you are at a competitive disadvantage,” Rayapati said, suggesting that unequal access to frontier AI systems could provide certain companies with a substantial competitive advantage over others.

The debate comes at a time when India’s technology sector is already confronting broader questions about how AI could reshape global employment patterns. Earlier this week, U.S.-based real estate technology company Opendoor announced the closure of its India office less than two years after expanding operations there. CEO Kaz Nejatian said the company wanted to move operational work closer to customers in the U.S. while also transitioning toward smaller AI-native teams.

Although Opendoor did not specify the extent to which AI-driven efficiencies influenced the decision, the move has fueled ongoing discussions about how advances in AI could reshape international technology workforces and what those changes could mean for India’s longstanding role as a global engineering hub.

Beyond Anthropic

Beyond the startup ecosystem and AI developers, Anthropic’s decision has also intensified broader discussions among Indian technology leaders regarding dependence on foreign AI infrastructure.

Sridhar Vembu, founder of the Indian software company Zoho, argued that the incident demonstrates that “technology is the ultimate weapon” and encouraged Indian organisations to adopt smaller, open-source AI models more aggressively.

“What can our government do right now? Ensure that orgs in India embrace smaller models, both Indian and Chinese open source ones,” Vembu wrote on X.

Investor and former Infosys executive Mohandas Pai responded to Vembu’s comments by arguing that the situation underscores the need for a significantly more ambitious national AI strategy. Pai urged the Indian government to dramatically increase investments in artificial intelligence, computing infrastructure, and deep technology development.

“We are way behind and need a national mission to get going quickly,” Pai wrote, proposing an annual ₹500 billion (approximately $5 billion) AI and deep technology fund alongside a ₹2 trillion (around $21 billion) credit guarantee program to support cloud infrastructure, semiconductor manufacturing, and hardware development.

Pai’s proposal would far exceed India’s current AI commitments. In 2024, the Indian government approved the IndiaAI Mission, allocating ₹103.72 billion (around $1.2 billion) over five years to expand computing infrastructure, support AI startups, and strengthen indigenous AI capabilities.

Despite growing momentum surrounding AI and continued government support for domestic innovation, it is a relatively modest participant in frontier AI model development. Only a small number of startups are actively developing building models, including Sarvam, which released open-source models earlier this year. Meanwhile, another prominent AI startup, Krutrim, shifted its focus toward cloud and AI infrastructure services after initially emphasising foundational model development.

Instead, much of India’s AI industry has focused on developing applications and specialised models built on existing foundation models. One recent example is Avataar AI, which introduced a video generation model earlier this week designed to offer a lower-cost alternative to competing products from Google’s Veo, Kling, Luma, and Runway.

Not everyone believes financial investment alone is the primary obstacle. Responding to Pai’s proposal, Lightspeed partner Hemant Mohapatra argued that the biggest barriers to building globally competitive AI companies are talent, access to computing resources, and strong execution rather than simply increasing funding commitments.

Mohapatra estimated that training a frontier AI model may require investments ranging from hundreds of millions to several billion dollars, depending on the underlying approach. However, he added that successful AI companies have historically gradually increased their capital requirements as adoption and commercial demand grew.

For several technology policy experts, however, the implications extend well beyond startups and AI developers.

Prasanto Roy, a New Delhi-based technology policy expert who advises multinational corporations, said the episode is likely to reinforce concerns within the Indian government regarding strategic technological autonomy. He compared the situation to the lessons many countries drew after Russia lost access to SWIFT and other components of the global financial system following its invasion of Ukraine.

Roy said the decision is likely to trigger a significant nationalist response in India and described Washington’s move as poorly judged, arguing that its consequences extend beyond Anthropic itself.

“Even if this is corrected or reversed, the Anthropic episode shows there’s no such thing as a geopolitically neutral foreign LLM,” Roy said. “American AI models are bound to American geopolitics.”