Beyond Shanghai: Why India must write its own AI rulebook

On July 16, in a Shanghai conference hall lit for a coronation, Xi Jinping stepped on stage to do something the People’s Republic of China has seldom attempted seriously before: not merely to build the world’s most-used artificial intelligence, but to write the rules that govern it. By the time the World Artificial Intelligence Conference (WAIC) closed, twenty-nine countries had put their names to a founding agreement establishing the World Artificial Intelligence Cooperation Organization — WAICO — an intergovernmental body headquartered in Shanghai.

Look closely at the guest list and the message becomes abundantly clear. Russia, Pakistan, Indonesia, Brazil, South Africa, Malaysia, Kazakhstan, Senegal, Cuba, Venezuela — roughly a dozen Asian and ten African states, a coalition consciously assembled from the Global South. Four of the five founding members of BRICS signed. The one that did not was India.

That absence of India is the most consequential thing about what happened in Shanghai. And it frames the defining technology-sovereignty question of the year for New Delhi: should India join Beijing’s AI rulebook, actively counter it, or write one of its own?

 The seduction is real

It is tempting to dismiss WAICO as choreography — another Chinese acronym destined for the graveyard of grand initiatives. That would be a mistake. Beijing has read the developing world’s grievance with precision, and it is selling the antidote.

“The development of artificial intelligence should not be a solo performance by any single country,” Xi told the delegates, “but rather a symphony of global cooperation.”

He went further, taking direct aim at Washington’s export controls: nations “should together oppose the practice of overstretching the concept of national security in the field of artificial intelligence.” As Fortune reported the scene, Beijing was offering an “olive branch” — and attaching gifts to it. China pledged five thousand AI training opportunities for developing countries over five years and free access to MAZU, its AI-powered meteorological early-warning system, for some thirty nations.

The subtext was unmistakable.

“China will not let America monopolise AI technology,” George Chen of The Asia Group told Fortune, reading Xi’s remarks as a promise of reliability to a Global South tired of being told what chips and models it may not have.

The economics reinforce the diplomacy.

While Silicon Valley locks its most powerful models behind subscriptions and Washington debates who may download them, China is doing the opposite — giving its models away. As Forbes columnist Craig S. Smith documented, when the United States moved to restrict a leading American model, a Chinese lab answered within days by open-sourcing a rival, and Washington reversed course within eighteen days.

“The champion of open markets reached for state control over who may use a private company’s product,” Smith wrote, “while the government most obsessed with centralised control is handing its most powerful technology to the world.” Zhipu’s founder, Jie Tang, put the pitch plainly: “Advanced AI should be open to anyone, not belong to a small group, or be pulled back at will.” For a finance ministry in Dakar or a start-up in Jakarta, that is not propaganda. It is a price list.

 The catch buried in the gift

Yet a rulebook is never neutral, and this one least of all.

Standards are the quiet infrastructure of power — whoever sets them decides what “safe”, “trusted”, and “compliant” AI will mean for the next generation, and whose systems clear the bar. WAICO is the institutional machinery to make those definitions permanent.

Beijing is already laying the pipework.

Alongside WAICO, it announced international AI cooperation centres linking the new body to ASEAN, the African Union, the Arab League, CELAC and the Shanghai Cooperation Organisation — the connective tissue of a parallel order. Writing for the Carnegie Endowment for International Peace, the governance scholar Arindrajit Basu warns that Beijing “views the governance of AI through the lens of entrenching state control, rather than safeguarding individual liberties” — and is now moving beyond exporting infrastructure towards “recrafting global rules, norms, and institutions of AI governance.”

Free models and training slots are the sweetener.

The main course is a worldview in which the state, not the citizen, sits at the centre of how machines see, sort and score human beings — one whose exported systems, Basu notes, “might automatically censor content embracing liberal norms such as the right to protest, regardless of where the model was deployed.”

This is where India’s instinct to stay out looks less like timidity and more like judgement.

A body whose founding charter was signed for Pakistan in Shanghai by its Deputy Prime Minister, Ishaq Dar, and convened to codify a state-centric model of technological control is not an easy room for New Delhi to enter on someone else’s terms. India’s former Permanent Representative to the UN warned that WAICO could end up defining the “new global AI order”, and that “India must be vigilant.”

Vigilance, here, began with declining the invitation.

 Two rulebooks, opposite foundations

The deeper reason New Delhi hesitates is that the two emerging rulebooks are built on incompatible foundations.

The Western-aligned tracks — the European Union’s risk-based AI Act, the network of AI Safety Institutes in the United States and Britain, the Global Partnership on Artificial Intelligence, which India itself has chaired — gate advanced systems behind safety testing, human-rights guardrails and values-aligned checkpoints. WAICO inverts the order of priorities.

It strips away political and ideological entry conditions and dangles cheap, open-weight models and shared infrastructure before any government that wants them, with state sovereignty rather than individual liberty as the organising principle.

For much of the Global South — weary of being lectured on values and locked out of compute — the second pitch is by far the easier sell. Which is exactly why the first cannot be India’s only reply.

 But abstention is not a policy

And this is the uncomfortable half of the calculus. Staying out keeps India clear of Chinese rule-making — but it also means the rules get drafted without an Indian pen in the room. As India prepares to chair the BRICS summit in September, it faces the awkward optics of leading a grouping in which every other founding member has already signed on to Beijing’s AI architecture.

The rhetorical high ground, too, is being conceded.

At the Shanghai gathering, the UN Secretary-General, António Guterres, warned that “technology that will shape the future of humanity must be shaped by all humanity” — the very language of inclusion India has spent years making its own, now amplified from a stage Beijing built.

An empty chair persuades no one in Nairobi or Hanoi. If the choice on offer is a generous Chinese “yes” or a silent Indian shrug, the Global South will take the models that are free today and worry about the rulebook tomorrow. Strategic autonomy that curdles into self-isolation is not sovereignty; it is irrelevance with better branding.

The Observer Research Foundation captured the emerging reality soberly: “most countries will engage with both ecosystems while balancing technology access against dependence, governance standards, and policy autonomy.” They will hedge. India’s task is to make sure it is one of the ecosystems they hedge towards — not a bystander in a two-horse race between Washington and Beijing.

 India has the third rulebook

Here is the good news New Delhi has been too modest to press: it is not starting from zero. In February, India hosted the AI Impact Summit at Bharat Mandapam and pitched, in the words of one account, a “third way” beyond the American and Chinese models — AI anchored in digital public infrastructure, the Aadhaar-and-UPI template that has already shown the developing world how to build population-scale digital systems it actually owns. Narendra Modi’s framing — “AI for All,” artificial intelligence as “a medium to enhance human values and potential” — is precisely the language a Global South audience wants to hear, minus the surveillance-state undertow.

That is a genuine third pole, and it is India’s to lose. To make it real, four moves are urgent.

First, ship a rulebook, not a slogan. India should convert the Summit’s goodwill into a standing Global South AI compact — a lightweight, treaty-adjacent body offering shared standards for safety, data governance and content authenticity, built on open protocols rather than a single vendor’s stack. Beijing has built a chartered institution; India cannot answer with another declaration.

Second, export the DPI advantage aggressively. The world’s most credible non-Western model of digital sovereignty already runs through Indian rails, and the domestic backbone is being built: the IndiaAI Mission is subsidising some 34,000 GPUs and underwriting a score of home-grown “sovereign” models designed for developmental needs rather than political control. Bundling India’s open models, compute-sharing and DPI playbook into a single offer to African and Southeast Asian governments answers China’s giveaway with something China cannot replicate: technology that leaves the recipient in control of its own data.

Third, refuse the binary in public. India need not join WAICO to engage its members. It can talk to Brasília and Jakarta bilaterally, keep its seat in the US-led camp that courts it, and still decline to sign anything that hands Beijing a veto over the definition of “trusted” AI. Non-alignment, updated for the age of the algorithm, is not fence-sitting — it is keeping the pen.

Fourth, use the BRICS chair; don’t apologise for it. Beijing is already working to thread WAICO’s standards into BRICS and the Shanghai Cooperation Organisation — forums where India sits as a full member — so New Delhi’s task is twofold: block consensus on restrictive Chinese mandates from within and turn September’s summit from a trap into a microphone. India should walk into its own gathering not defending an absence but tabling an alternative — and inviting the very Global South Beijing is courting to help write it.

 The call

China has done India an inadvertent favour. By moving first and moving boldly, it has ended the era in which New Delhi could treat AI governance as a seminar topic. The rulebook is being written now, in Shanghai, with or without an Indian signature.

The answer is not to sign, and not to sulk.

It is to publish a better rulebook — open where Beijing is closed, citizen-centred where Beijing is state-centred, and offered to the Global South not as a leash but as a ladder. India has the credibility, the technology stack and the story to do it.

What it needs is the nerve to stop reacting to other people’s summits and start convening its own.

Sovereignty in the age of AI will not be won by owning the biggest model. It will be won by whoever the rest of the world trusts to write the rules. That contest has just begun — and India, for now, is choosing to watch. It should choose, instead, to lead.

The pen that will write the world’s AI rules is, for a moment, still unclaimed.

Beijing has lunged for it; Washington guards its own. India’s task is neither to grab nor to grovel, but to offer the Global South a steadier hand, and to do it before the ink of Shanghai dries into permanence.

(Akhileshwar Sahay is a multidisciplinary thought leader. He is a keen watcher of the changing national and international scenario. Views expressed in the above piece are personal and solely those of the author. They do not necessarily reflect Firstpost’s views.)