Why the world may need an AI stewardship treaty

Should a cabal of countries and corporations become permanent custodians of the world's most powerful intelligence?

The world needs an AI stewardship treaty

Artificial intelligence is rapidly becoming a foundational capability, one that will shape economies, militaries, scientific discovery, healthcare, education and governance. The countries that master it will enjoy enormous strategic advantages over those that merely consume it.

As governments and technology companies debate how frontier AI should be governed, one question remains curiously absent from the conversation: who gets to decide which countries are trusted with humanity's most powerful intelligence?

The question is usually framed as one of safety, but that is only one part of the story.

Should a small group of countries and corporations become the permanent custodians of the world's most powerful intelligence, while everyone else remains dependent upon them?

Every generation has witnessed transformative technologies becoming instruments of geopolitical power. Nuclear technology, advanced semiconductors, satellite systems and cryptography have all become subjects of export controls, strategic alliances and technological denial. Artificial intelligence is rapidly joining that list.

Who determines where the technological frontier should stop, and for whom? Too often, this debate proceeds from the unstated premise that today's technological powers are naturally the world's most trustworthy stewards. Why should the rest of the world simply accept that proposition?

The post-war international order has produced extraordinary scientific and economic progress. Yet it has also witnessed repeated interventions by major powers in pursuit of their own strategic interests.

Recognising this reality is not an argument against any particular country. It is an argument against concentrating unprecedented technological power in the hands of a few. This matters because AI differs fundamentally from previous strategic technologies.

AI is becoming the cognitive infrastructure upon which economies, governments and societies will increasingly depend. The same foundation models that accelerate scientific discovery can improve healthcare, education, manufacturing, agriculture and climate resilience.

Restricting access to frontier AI therefore shapes far more than national security. It determines who can innovate, who can compete, who creates economic value, and ultimately who prospers.

IMPLICATIONS FOR INDIA

For countries such as India, the implications are profound. India is no longer merely a consumer of global technology. It possesses world-class engineering talent, globally admired digital public infrastructure, a thriving start-up ecosystem and growing frontier research capabilities. Yet if access to the most capable AI systems remains controlled by a handful of foreign governments or corporations, India and many other capable nations risk becoming permanently dependent on technologies that will underpin every sector of their economies.

None of this implies that frontier AI should simply become freely available to everyone. Legitimate security concerns must be addressed. The world does need robust safeguards against catastrophic misuse.

But preventing misuse is fundamentally different from preventing capability. If AI is becoming global public infrastructure rather than merely another strategic technology, then its governance cannot be modelled solely on arms-control regimes. It must also address questions of equitable access, institutional legitimacy and shared stewardship.

The appropriate precedent may therefore not be a single treaty but a hybrid framework.

One element could draw from the logic of the Chemical Weapons Convention: establishing internationally agreed prohibitions against clearly harmful applications of artificial intelligence, together with verification mechanisms and independent oversight. Another could borrow from international regimes governing domains of shared global importance such as the Law of the Sea and the Outer Space Treaty, which recognise that resources of profound significance require cooperative governance rather than exclusive control by a small number of states.

Such a framework would distinguish between restricting malicious uses of AI and denying responsible nations access to frontier capabilities. Safety, equity, accountability and international legitimacy would become complementary objectives rather than competing ones.

A genuinely multipolar AI ecosystem, involving the United States, Europe, India, China and other capable innovation centres, may ultimately prove more stable and equitable.

Perhaps what the world actually needs is an AI Stewardship Treaty: one that prohibits harmful uses of artificial intelligence while ensuring that its transformative benefits remain safely, equitably and legitimately accessible to all responsible nations.

Originally published as an Op-ed in The Hindu BL on August 17, 2026.