Anthropic Ban Sparks India Inc.’s Call for Sovereign AI

Anthropic last week disabled access to its new Fable 5 and Mythos 5 for all customers following a US government directive. The move has drawn sharp reactions across the world, and experts believe it will have an impact on India Inc. as well.

For the uninitiated, Fable 5 and Mythos 5 are two iterations of what Anthropic described as the state-of-the-art “Mythos-class” AI model. Anthropic’s cybersecurity focused AI model, Mythos, has already sparked widespread concerns about its potential to be misused for automated sophisticated cyberattacks and preparedness of enterprises around the world. 

While Anthropic has tried to gatekeep Mythos with select access, the latest move raises more questions about such AI models, and their own integrity. 

Heart of the matter

According to Anthropic, the US government has issued an export control directive to suspend access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the US, citing national security authorities.

“Our understanding is that the government believes it has become aware of a method of bypassing, or “jailbreaking” Fable 5. We reviewed a demonstration of this specific technique being used to identify a small number of previously known, minor vulnerabilities. These vulnerabilities all appear relatively simple, and we have found that other publicly-available models are able to discover them as well without requiring a bypass,” the company said in a post.

“..We are complying with the government’s legal directive and are removing access to Fable 5 and Mythos 5 for all users. However, we disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people. If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers,” it further said in the post. 

The move has already drawn sharp reactions in the tech space, especially in India. 

For instance, Chairman of Aarin Capital and Manipal Global Education Services T V Mohandas Pai sought earmarking of INR 50,000 crore to support deep tech and AI development in India.

Zoho founder Sridhar Vembu noted: “Technology is the ultimate  weapon. National sovereignty, national security, all of it is now about technology… Globalization is dead and Bharat must find her own way ahead.”

“What can our government do right now? Ensure that orgs in India embrace smaller models, both Indian and Chinese open source ones. With a bit of effort, we can make them work. Anyway, why pay money to people who don’t even want to sell to you?

We must deepen our R&D. Sarvam has been on it and we have been on it but remember that the latest models cost not only huge GPU budgets to train, the GPUs themselves are restricted. So we can’t afford the scale of money (of the order of $100+ billion to even get in the game!) and even if we could come up with the money, we can’t get all the GPUs. I would not like to ask the government to fund tens of billions of dollars on this anyway –  the money has far better uses…” he added in a post. 

Manoj Dhanda, Founder & CEO of Utho Cloud, tells Entrepreneur India that Anthropic’s move is about the US acting in their national interest, and India must learn from it. 

“That is the line our country has to wake up to. Owning your data is not the same as owning your capability. We can keep every byte inside India and still not control whether the technology our economy runs on works tomorrow morning. Think about what now depends on foreign AI in this country. Our banks. Our hospitals. Our startups burning runway. Our government departments. Our defence and research. All of it increasingly built on a handful of models owned by companies in another country, governed by a government that is not ours, that can withdraw access on a Friday evening for reasons it does not even have to explain to us,” Dhanda said.

He also noted that India has the talent, the engineers, the data centres and the demand to build a real sovereign stack. 

“Compute we control. A country that does not own its AI does not own its future. It rents it. And rent can be cancelled. So the question for all of us is simple. If the AI our country depends on was switched off tomorrow by a government that is not ours, what exactly is our plan? Dhanda added.

Has gatekeeping ever worked?

With the US sitting on the switch on-off button for the likes of Mythos, it’s important to understand can such technologies be actually guarded in letter and spirit. And such a move could actually fast track development of equivalent technologies in different corners of the world. 

History suggests gatekeeping a software capability rarely holds. Strong encryption was export-controlled as a munition in the 1990s, and within a decade the controls were largely unwound because the capability spread regardless. Frontier AI will follow a similar arc, with one difference: the predictable response to being gated is that the gated build their own. 

Ashish Tandon, Founder & CEO, Indusface, notes that the most lasting effect of last week’s restriction will not be containment. It will be an acceleration, as every capable nation now treats indigenous frontier AI as a necessity rather than an ambition.

“India is unusually well placed to answer that call, with one of the world’s deepest pools of engineering and AI talent already building at the frontier inside global labs. This is the moment the conversation moves from long-term aspiration to near-term investment. Restricting access does not slow the technology. It redistributes who builds it,” he added.

Though, one must not confuse between guardrails and gatekeeping. 

Prateek Bhajanka, CEO at Field CISO Advisory and former Gartner Analyst, tells Entrepreneur India that guardrails by Anthropic/ frontier AI model provider is an absolute necessity for the safe use of the power of AI and models as the intent of the use of the models cannot be predicted at the time of sign up.

Gatekeeping, on the other hand, is restricting access to the technology even after the guardrails for safe use have been implemented. The gatekeeping of a technology because of geo political reasons will not be conducive to the global technology ecosystem as it will erode the trust on global platforms as the access can be restricted any time.

“Fortunately, there are numerous open-source models which are coming close to the cut throat capabilities of Anthropic. India should become independent and work towards developing its own Sovereign AI to reduce any future economic risk arising from gatekeeping of technology,” Bhajanka explained.

The Countermeasure 

It appears that there is wide consensus that India must act now on improving its cybersecurity regime, especially in the era of Mythos and AI. But what exactly must it do?

The disadvantage is real, but the mitigation is not to wait for access to be restored. It is to remove single points of failure. Indian enterprises learned last week that a capability sourced entirely from one foreign provider can vanish in ninety minutes, so the first move is architectural: treat models as interchangeable, combine open-weight models you can run yourself with closed models where they genuinely lead, and never let a capability you cannot control become load-bearing. 

The companies that weather the next disruption will be the ones that depend on their own data and architecture, not on any single vendor’s model. The second move is collective. India Inc. should back indigenous frontier efforts with capital and demand, because a domestic fallback at the frontier is now a continuity requirement, not a point of national pride.

Bhajanka adds  that India’s long-term strategy should focus on normalizing the use of AI vs AI. The current practices are more manual rather than AI first. 

There are encouraging signs. CERT-in acknowledged the onslaught of vulnerabilities and issued an advisory to patch the critical vulnerabilities in internet facing systems within 24 hours. They also recommended patch automation to reduce the processing time. 

However, to stay ahead, organizations need to move faster in adopting AI- powered detection, response, and remediation capabilities. Building AI-native cybersecurity frameworks, investing in automation, and strengthening public-private collaboration will be critical to keeping pace with the evolving threat landscape.

Tandon of Indusface notes that India is not yet on the right track, and the gap is one of speed rather than intent. The same models being restricted can find exploitable vulnerabilities in widely used software faster than any human team can patch them. Analysts have specifically flagged banking, with its complex and often decades-old systems, as exposed. India’s critical sectors run on exactly that kind of infrastructure. 

“The regulatory conversation is moving, but it is calibrated to a threat timeline that no longer exists. The immediate action is to assume attackers now operate at machine speed and to build defences that close the gap between disclosure and exploit automatically. Long term, India needs indigenous capability at the frontier so its defenders are never dependent on tools a foreign directive can withdraw,” he added.