Mythos challenge: Delhi needs a say in AI regulation

Considering that India's numbers make it one of the biggest markets for AI models and that it is consumer data that greases these models, it must have a seat at the high table of global AI governance

The development of AI models such as Anthropic’s Mythos clearly shows that the pace at which artificial intelligence capabilities are being designed has now reached levels where not only are explosive new opportunities possible, but the likelihood of greater risks is clear and present. Mythos is said to be able to identify and exploit gaps or flaws in IT systems — the “zero day” vulnerabilities in critical infrastructure which even developers and firms are oblivious to. According to the UK’s AI Security Institute, it can “execute multi-stage attacks on vulnerable networks and discover and exploit vulnerabilities autonomously”. The threat it poses is obvious, especially when there have been reports of unauthorised access to it. This raises questions that centre not only on the development of such models, but also on deployment, transparency and accountability. Clearly, AI governance is no longer a seminar talking point.

On its part, Anthropic appears to be moving ahead carefully. It has said that around 50 companies — these include the likes of Google, Microsoft, Amazon and JPMorgan Chase — will get access to the model in order to assist them in fortifying their systems against the model’s ability to exploit their weaknesses. This step may be well intentioned — the results of this exercise are expected to be shared. But surely it must be asked: Can the decision of who gets early access to such disruptive models be left to a single company or individual? Or, in future, to the heads of leading AI companies as other models are developed? What happens when China releases its own Mythos? And who decides how these models will subsequently be deployed? After all, the incentive structures of the individuals will shape their actions. In technologies of such nature — which even non-state actors can use to try to exploit weaknesses — deployment must be preceded by oversight and a regulation that cuts across borders.

For India, the challenges are clear. On Thursday, finance minister Nirmala Sitharaman held a meeting to gauge the risks to the Indian financial system from Mythos. While this is certainly a crucial first step, the government must be mindful that the risks could well extend beyond the banking system to other critical sectors. Addressing these will require a concerted and coordinated effort at the global level. Considering that India’s numbers make it one of the world’s biggest markets for AI models, and that it is consumer data that greases these models, it must have a seat at the high table of global AI governance. It has to be able to shape the regulatory framework. In these fraught times of war and tariff, consensus is elusive but the good news is that Mythos has rattled all governments. This is reason enough for all to come to the negotiating table to work on a global governance architecture — and India should have a seat there.