AI is now a great-power game. Three difficult questions for India

Why in the News

The United States and China are opening a formal bilateral dialogue on artificial intelligence (AI) and its risks, with the Chinese President received in Washington by the US President on Thursday for their third summit in less than 10 months. The meeting comes barely 10 days after the Delhi BRICS summit, where China offered its own vision of international cooperation on AI. The two events set out opposite agendas for the same technology, one built on openness and diffusion and the other on preventing accidents, miscalculation and misuse. BRICS rhetoric on AI may be multipolar. Technological power at the frontier remains concentrated in the United States and China.

What is strategic stability in the AI age?

1. Strategic stability: A condition in which neither of two rival powers has an incentive to strike first or to race, sustained by shared understanding of each other’s capabilities and by channels that stop an accident being read as an attack.
2. Origin of the term: It comes from nuclear arms control, where the United States and the Soviet Union negotiated rules to reduce mutual risk during an active contest for advantage.
3. What it covers in AI: Preventing accidents, miscalculation and misuse arising from the most capable models, including their military applications.
4. What it does not cover: It is narrower than agreement on the technological order as a whole, so a strategic stability understanding can be reached with chips, export controls and market access still contested.

What does China’s BRICS offer on AI amount to?

1. Openness, access and diffusion: The Chinese framing at Delhi placed the emphasis on spreading AI capability rather than on containing its risks.
2. World Artificial Intelligence Cooperation Organisation (WAICO): BRICS members were invited to join this body, which China launched in Shanghai this summer as a venue for international AI cooperation.
3. Open models for the forum: China proposed that BRICS host an open-source AI community, which makes models and tools available for others to use and adapt.
4. Champion of affordable access: China presented itself as the advocate of developing countries seeking affordable access to AI models and applications.

Why does the Washington agenda invert the BRICS agenda?

1. Subject of the conversation: The Washington discussion concerns the dangers generated by increasingly powerful models and the need to prevent accidents, miscalculation and misuse, which is the opposite end of the technology from access and diffusion.
2. Who is writing the security rules: As the world’s leading AI powers, the United States and China are beginning to shape the security and geopolitical architecture of the new technology.
3. The distance between rhetoric and capability: Multipolar language at BRICS does not change where the most advanced capability sits, and the journey from Delhi to Washington in barely 10 days is what makes that visible.
4. Cold War precedent: The United States and the Soviet Union competed and cooperated in the nuclear and space domains at the same time, and the rules they negotiated inevitably affected the rest of the world.

What has the US-China AI channel produced, and where does it still diverge?

1. Opening of the channel: A tentative channel was opened by the Biden administration after the two leaders met at Woodside, California, in November 2023.
2. The one concrete outcome: At Lima a year later the two leaders affirmed that humans, not artificial intelligence, must control decisions on the use of nuclear weapons.
3. Guardrails discussed without result: The two leaders discussed possible AI “guardrails” when they met in Beijing in May, and no formal arrangement followed.
4. What is now on the table: After talks in New York last weekend between the US Treasury Secretary and the Chinese Vice-Premier, Washington said the two sides had considered a standing AI dialogue and a notification system for incidents with national-security consequences.
5. Why such a mechanism matters: In a crisis the first requirement is not agreement on the entire technological order. It is the ability to establish what has happened, whether it was accidental or deliberate, and who is responsible.
6. The American objective: The United States wants a framework to manage the risks posed by powerful models and their military applications.
7. The Chinese objection: China treats safety discussions as a possible instrument for preserving American technological advantage and for legitimising restrictions on its own access to advanced computing.
8. The named disagreements: The two also disagree on open-source models, data governance, state regulation and the legitimacy of export controls.
9. Entanglement with trade: Safety discussion under the present US administration runs alongside disputes over advanced chips, export controls, rare earths, tariffs, computing power and the race to build the most capable models, so no grand bargain is at hand and what is sought is a narrower understanding on strategic stability.

What three questions does this leave India?

1. Navigating the technological contest: India sits inside the US-led Pax Silica initiative on technology and supply chains, which follows from its deep integration with the American digital and AI ecosystem, and it has stayed out of China’s WAICO amid enduring security problems and mutual distrust. As a BRICS member it still cannot avoid AI rhetoric shaped by the forum’s dominant power.
2. Competing offers to developing societies: India’s emphasis on making AI work for developing societies found considerable resonance at the Delhi AI summit earlier this year. China’s deeper capabilities in computing, infrastructure, hardware and advanced models let it offer the Global South a wider package of technological cooperation.
3. Rule-setting by two powers: A dialogue that begins on strategic stability could become an effort to set rules for everyone else, producing discriminatory technology controls against states outside it.
   Counter-pressure: Geopolitical rivalry and competition for markets between the two powers could limit such collusion and create leverage for emerging technological powers such as India.

Challenges to a US-China understanding on AI risk

1. Absence of a countable unit: Model capability cannot be counted or inspected the way delivery systems were, so compliance with any understanding cannot be confirmed. Eg. The New START treaty rested on declared launcher numbers and on-site inspection, and neither has an analogue for a training run or a set of model weights.
   The Fix: Anchor the dialogue on physical proxies such as data-centre compute capacity and chip transfers, which are countable and already tracked.
2. Controls that leak through third countries: Export controls bind firms rather than territory, so restricted chips move through intermediaries in jurisdictions outside any bilateral arrangement. Eg. Washington’s October 2022 controls on advanced computing chips were widened a year later after routes through third countries emerged.
   The Fix: Tie the notification commitment to a shared list of controlled items, so a transfer is traceable rather than disputed after the fact.
3. Capability held by private developers: The most capable models are built by companies, so a state-to-state understanding does not bind the actors doing the building. Eg. The 2023 Bletchley Park declaration on frontier model safety rested on voluntary commitments from developers rather than obligations on them.
   The Fix: Require incident reporting from developers under domestic law, so a state can honour a notification commitment over conduct it does not itself control.
4. Exclusion of the states the rules will govern: Terms settled between two powers become conditions of access for everyone else without their participation. Eg. India stayed outside the Nuclear Non-Proliferation Treaty and needed a separate Nuclear Suppliers Group waiver in 2008 before it could trade in nuclear material.
   The Fix: Route AI risk rules through a body with wider membership, so conditions of access are negotiated by the states they bind.

Conclusion

Two states are writing the safety rules for a technology they also lead, and the rules will reach states that were not in the room. India’s exposure is therefore not the rhetoric of any forum but the possibility of inheriting terms it did not negotiate. What decides its position is measurable and domestic: computing power, semiconductors, models, research, capital and talent. The thing to watch is whether the risk channel the two powers settle on is opened to third states or kept strictly bilateral.

What is technology diplomacy?

1. About: The use of a state’s technological assets, innovation ecosystem and digital standards to build international partnerships, influence global governance and secure national interests.
2. Two halves: Tech-cooperation shares working tools such as digital payment systems with other states. Tech-governance shapes the norms for artificial intelligence and 6G.
3. Named pillars: Exporting digital public infrastructure as a public good, securing lithium and cobalt through the Mineral Security Partnership, building trusted semiconductor supply chains, and cyber diplomacy.
4. Why it is a distinct instrument: A state that sets a standard shapes the conduct of every state that adopts it, which gives technological capability a diplomatic return that trade or aid does not.

Government Initiatives on artificial intelligence

1. IndiaAI Mission (2024): Approved with an outlay of Rs 10,371 crore and implemented by IndiaAI under the Ministry of Electronics and Information Technology (MeitY), it builds compute, datasets, models, skills and safety capacity for startups, researchers and students.
2. IndiaAI Compute: A national AI compute grid of over 38,000 graphics processing units, offering eligible users up to 40 per cent lower compute costs.
3. AIKosh: The national AI dataset repository, carrying over 3,000 datasets and 243 models across 20 sectors for developers who cannot assemble their own data.
4. IndiaAI Safety Institute: The mission’s trust pillar, covering bias mitigation, privacy, explainability and AI governance.

Back2Basics: Pax Silica

1. What it is: A United States-led initiative on technology and supply chains, which India is part of.
2. Purpose: Building resilient semiconductor and electronics supply chains among trusted partners, so production is not concentrated in a single jurisdiction.
3. Why it matters for India: Membership follows from India’s integration with the American digital and AI ecosystem, and it sits opposite the China-led venue India has declined to join.

Matching Previous Year Question

“[2026, GS2, 10 marks] “BRICS acts as a powerful counterweight in global governance, actively amplifying the voice and influence of the Global South.” Explain the role of BRICS in projecting itself as an alternative to other groupings.”