The 2026 AI Impact Summit in India is the latest iteration of an event that has become a bellwether for global discourse around the AI industry and if/how it can be governed. Each edition (the first hosted by the UK; the second, virtually, by South Korea; and the most recent one by France, with India as cohost) has been a complex tussle between the aspirations and realpolitik of the host countries and the AI industry. Originally imagined as a “safety” summit, that vision has diluted over time, shifting focus from safety to action and, now, to impact. The summits have enjoyed participation from tech company CEOs, making them frenzied media events—even as other global governance fora are studiously ignored by the same industry leaders. Typically organized as invite-only gatherings, with a mix of curated discussions and private roundtables, the summits are sites of hot contestation around who gets heard. For instance, Day 2 has consistently been set aside for closed-door government-industry conversations, with a vanishingly small minority of civil-society actors at the table.

While these summits don’t feed into traditional multilateral processes, they’ve been celebrated for eliciting consensus statements from governments around the risks posed by advanced AI (the Bletchley Declaration) and securing high-level commitments from the tech industry (the Frontier AI Safety Commitments at Seoul). But at a time when we need robust, enforceable regulation to hold the AI sector to account, the legitimation of voluntary regimes can create the illusion of progress while leaving companies to continue business as usual. Moreover, these formal undertakings address a narrow band of future-looking risks from the most advanced AI systems, and are much quieter on burgeoning social and economic risks from AI as they exist today, from labor to competition to children’s safety online. Finally, the summits can easily seem disconnected from current geopolitical realities, with their misplaced lip service to multilateralism and collective solutions at a time when rogue states and rising authoritarianism are actively eroding the very foundations of global cooperation. 

For civil society engaged in holding tech power to account, the 2026 AI Impact Summit raises a fundamental strategic question: Should we even engage? At a time when fights to challenge the unaccountable AI industry and its scale-at-all-costs trajectory are most potent at the national (and, with data centers, hyperlocal) level, are these summits a distraction from more meaningful organising? The combination of organizational opacity, often box-checking participatory channels, and an agenda that bends erratically to the will of individual state and corporate actors means we have plenty of reason to be skeptical. 

But the discursive terrain opened up by this year’s Summit may offer opportunities to challenge dominant framings—and to lay claim to alternative ones. The UK Safety Summit, for example, foregrounded existential future risks related to AI (from bioweapons to killer robots), yet it ended up spurring heated debate around the need to refocus attention on existing material harms. The French Action Summit, in turn, centered on a public-interest-oriented theme that offered fertile ground for affirmative alternative imaginations outside of industry-led AI. Many in civil society advocated for a concomitant focus on competition, arguing that efforts to catalyze an open public-interest AI ecosystem would fail if not accompanied by a robust commitment to addressing the major distortions in today’s AI market, rigged in favor of tech giants. Although the summits offer a narrow band of opportunity to civil society, we should be clear-eyed that this is not a level playing field, made worse by an increasingly fragmented information environment that’s often hard to pierce through: despite the formal importance of public-interest AI for the French Summit, its lasting echoes in the news cycle were of President Macron chest-thumping about France’s champions in the AI race.

So, as we look to the New Delhi Impact Summit, now a month away, it’s worth doing early forensics on the themes that are likely to be instrumental. The current framing of themes (People, Power, and Planet as the three “sutras”) and subthemes (“chakras”) across the official agenda may be intentionally broad but can appear unfocused. India has the hard task of organising a global summit in a fractious geopolitical environment, where it is attempting to balance multiple (often competing) objectives : projecting itself as a leader of a new form of technological non-alignment anchored in strategic autonomy, while remaining dependent on Big Tech for critical infrastructure and innovation. This tension is further compounded by the increasing volatility of US- India relations, and the need to contend with China’s role in global AI governance. 

With this context,an examination of official Summit materials and related discourse reveals two underlying threads that will be pivotal to understanding the Indian government’s unique posturing and aspirations: Global South leadership; and the notion of a third way. 

Global South Leadership on AI: Piercing Through the Hype

India is the first Global South country to host the AI Summit. Official messaging emphasizes the Summit as an opportunity to “give voice to the Global South” and democratize AI resources for all. In many ways, this crystallizes the country’s steady pursuit of economic and diplomatic leadership in technology and development. India’s G20 presidency (2023) is widely heralded as strengthening India’s role as a bridge between the Global North and Global South, and this AI Summit is framed as an extension of that positioning. 

“Global South” is routinely invoked as a coherent political subject that implies not only homogeneity, but also underdevelopment and technological illiteracy. The “Global South,” of course, represents a diverse group of countries—and India isn’t the only one vying for leadership, investments, and a seat at the table. While India has been positioning itself as “data rich” and the “use case capital” of the world, other countries like Rwanda and Nigeria have positioned themselves as sites for “scaling hubs,” and the UAE is an increasingly attractive site for Big Tech to raise the capital it needs to finance AI infrastructure build-out. Elsewhere, we have seen massive AI for social good or AI for development plays. 

Across the angles, the upshot is the same: Global South countries are advertising their populations as a path to scale for AI companies. By offering their wide base of customers and data to Global North corporations, these countries hope to demonstrate they are investment-worthy. They also offer the feel-good narrative of using AI to improve the lives of impoverished populations. 

The prevailing sentiment across many Global South countries is that failure to become active users and innovators of AI will result in further marginalization. At the same time, there is an almost utopian belief that AI can resolve long-standing structural problems that have persisted for decades, if not centuries, from poverty to climate crisis. This dual framing, AI as both a developmental necessity and a competitive imperative, has generated significant enthusiasm and urgency.

However, the promise of “AI for good” or “AI for development” closely resembles earlier development narratives: it obscures trade-offs, externalities, and power asymmetries. There is little transparency around who bears the costs, who captures the value, and whose priorities ultimately shape these technological pathways. Under the veneer of Global South solidarity lie divergent interests that obscure the darker picture of extractivism. Global South labor powers AI via content moderation, data labeling, and even humans masquerading as AI. Global South countries have critical minerals used throughout the AI supply chain. Land, energy and water in an already resource-strapped Global South are increasingly being used for data centers in the never-ending quest for more resources. Unequal dynamics are structuring not only relations between Global North and Global South, but also relations between and within countries in the Global South, where India for example is actively exporting software platforms and services to other Global South countries. 

So as we look ahead to the Summit, we must be ready to scrutinize claims to Global South leadership, and themes that pay lip service to the idea of geographical diversity and empowerment without contending with the power asymmetries and extractive dynamics underpinning the current market structure. For example, will the spotlights on “linguistic diversity” be in service of making dominant LLM products accessible and legible to more populations—or is there support for genuinely localized alternatives. Similarly, we need to scrutinize AI for development agendas that read like sales pitches for government adoption of AI, especially with “AI for government” strategies from OpenAI, Google, and others ready for uptake.

The Promise of a “Third Way”: Substance or symbolism?

In its claim to Global South leadership, India has positioned itself as a “third way,” an alternative to US and Chinese approaches that mobilizes the promise of using technology to benefit not corporations or the state but the public. This people-centered framing is appealing, and one that many civil society and philanthropic actors increasingly champion as being necessary to reclaim “AI sovereignty” from exclusively industry or geopolitical imperatives. 

India’s approach to the “third way” is best captured in its global push for digital public infrastructure (DPI), a buzzy shorthand for a state-promoted technology stack modeled on India’s digital identity (Aadhaar), digital payments (UPI), and data exchange systems. DPI’s supposed success, highlighted during the Indian G20 presidency, promises a template for building scalable, context-specific, and cost-effective tech solutions, particularly for developing countries seeking alternatives to Big Tech-dominated systems. DPI is being linked to AI as well, though it is still unclear what this means in practice.

Lessons from DPI to AI are cautionary tales of the risks of misusing a “public-interest” narrative. DPI promoters use narrow technical precepts (like Open APIs) to make bold claims about openness, which is imagined as automatically benefitting the public. In practice, many of these applications have been experienced as closed, inscrutable systems that enable surveillance and facilitate private capture of public functions at an enormous human cost. The use of algorithmic decision-making for technology to mediate access to welfare, for example, has locked people out of benefits and other critical services—with little accountability. And despite messaging around challenging Big Tech hegemony, India’s open protocol for payments, for example, is today dominated by Google Pay and Walmart-owned PhonePe. 

Similar debates around “openwashing” are very much alive in the AI context, where flimsy definitions (is Meta LLaMA really open-source AI?) function to obscure the infrastructural dependencies that run through the AI stack. We should be wary, in this context, of the India Summit discourse around inclusion for social empowerment and the creation of an AI commons that does not engage with the use of “openness” as a hegemonizing tactic by Big Tech.

Another undercurrent of this approach is that governance is no longer really the government’s problem. It instead swings between a techno-legal approach, where code is law, and voluntary self-regulation, where rules are not enforceable. The “Practical Guidelines” in India’s recently-released AI Governance Guidelines ask regulators to “support innovation while mitigating real harms; avoid compliance-heavy regimes; promote techno-legal approaches; ensure frameworks are flexible and subject to periodic review.” Taken together, these recommendations suggest an effort to keep AI governance depoliticised, adaptable and innovation-friendly, reflecting a broader emphasis on maintaining India’s global competitiveness.

The summit track on “democratization” of AI has potential, and could lead with a challenge to the current concentration of compute and data resources in a few Big Tech hands, but could just as easily sidestep questions of power distribution in favor of a more benign “access” mandate (a.k.a. more compute, and everyone gets an LLM). India’s “use case capital” positioning in the lead-up to the Summit blends in well with this posture, allowing the country to focus on a burgeoning downstream AI start-up ecosystem while sidestepping a key question of value distribution and Big Tech power: Will these start-ups be more than barnacles on the hull of Big Tech companies that still control the underlying computing and foundation model infrastructure? This concern is reinforced by a recent wave of industry collaborations that shows that India is open for Big Tech business—despite its public emphasis on “homegrown” products. India’s OpenAI Learning Accelerator emphasizes expanding access to AI for educators; Anthropic’s India strategy is focused on adoption of AI in agriculture, education, and the availability of the technology in Indian languages; Google is pushing DPI and AI integration in healthcare, and using India as a resource for model testing, fine-tuning, and improvement. Meta, Microsoft, and Amazon have made similar announcements on cloud infrastructure, skilling, and application layer growth. These strategies effectively render India as a site for localizing existing models and scale, but not as a locus of control or leadership. 

It is sovereignty, then, that forms the ever-present but unspoken background of the Summit. In courting US tech capital, India’s own state-centric aspirations of digital sovereignty do not find utterance in official Summit documentation, even as Big Tech itself has cashed in, offering “sovereignty-as-a-service” to governments. As we’ve learned from the concrete experiences of India’s “third way”, in the tussle between states and Big Tech, it is always regional, community-oriented, bottom-up notions of sovereignty that lose out. How, then, can this Summit genuinely be the Global South’s moment for challenging power in AI ecosystems? 

Stitching Together a Genuine “Third Way”: The Reframing Impact Series

There is wide-ranging global consensus that the unprecedented levels of unaccountable power in the AI industry are a key challenge of our time, from governments increasingly anxious about their core digital infrastructure becoming beholden to the whims of foreign tech CEOs to a general public that is grappling with the ever growing harms of the AI boom. 

In this context, the 2026 Summit, with its impact-oriented framing and calls to internationalism and Global South leadership, offers fertile terrain to build resistance to the status quo, and to stitch together the burgeoning national and local efforts that collectively represent a people-centered alternative But, as we unravel in this introductory essay, we expect potent themes like democratization, climate, and development to be presented in a defanged form—functioning to maintain dependence on prevailing power structures rather than challenge them. 

This is where our upcoming Reframing Impact: AI Summit 2026 series takes off. In this series, a partnership between AI Now Institute, Aapti Institute, and The Maybe, we draw together a wide-ranging network of advocates, builders, and thinkers from many parts of the world to bring attention to the limitations of the current discourse and move into the conversation we want to have.

We’ll be in conversation with Audrey Tang (Democratization), Naomi Klein (Climate), Abeba Birhane (AI for Good), Karen Hao (Data Rich), Joan Kinuya (Human Capital), Usha Ramanathan (DPI for AI), Meredith Whittaker (Open Source), Chinasa Okolo (Multilateralism), Kailash Nadh (Democratization), Timnit Gebru (Frugal AI), Nikhil Dey (Accountability) and more…