India Hosts First Global South AI Summit as Safety Researchers Sound the Alarm

The marble halls of Bharat Mandapam in central New Delhi hummed with last-minute preparations on Monday morning as India readied itself to host the largest artificial intelligence gathering the world has seen — and the first ever staged in a developing country. Prime Minister Narendra Modi was set to inaugurate the India AI Impact Expo at 5 p.m. local time, kicking off a five-day programme that organisers say will draw 250,000 visitors, 20 national leaders, and 45 ministerial-level delegations to the Indian capital .

The India AI Impact Summit 2026, running February 16–20, is the fourth annual global convening on AI governance after previous editions in Bletchley Park (2023), Seoul (2024), and Paris (2025). But its scope dwarfs anything that came before. Themed around three "sutras" — people, progress, planet — the summit aims to produce what organisers call a shared roadmap for global AI governance and collaboration . Whether that roadmap will amount to more than aspirational language is the question hanging over every panel and bilateral meeting on the schedule.

The guest list reads like a who's who of global power and Silicon Valley wealth. Sam Altman of OpenAI, Google CEO Sundar Pichai, and Anthropic's Dario Amodei have confirmed attendance, alongside India's own Mukesh Ambani . French President Emmanuel Macron and Brazilian President Luiz Inácio Lula da Silva are among the heads of state expected. Neither Donald Trump nor Xi Jinping will attend in person, though both Washington and Beijing are sending senior tech-policy officials — a calibrated diplomatic signal that the world's two AI superpowers want a seat at the table without committing their leaders to any binding outcome .

One notable absence has drawn particular attention. Nvidia CEO Jensen Huang, whose company's chips underpin the vast majority of AI training infrastructure worldwide, cancelled his planned appearance on Saturday. Nvidia said he was unable to travel to India at this time due to unforeseen circumstances, and that a senior delegation led by executive vice president Jay Puri would attend in his place . The cancellation sparked immediate speculation about whether the decision was health-related, political, or simply logistical — Nvidia declined to elaborate. For a summit billing itself as the definitive gathering of AI's most influential figures, the absence of the man whose hardware makes generative AI possible is conspicuous.

The summit opens against a backdrop of mounting unease within the AI industry itself. In the weeks leading up to the New Delhi event, a string of high-profile safety researchers have walked away from leading AI companies, publicly warning that the technology is advancing faster than the guardrails being built around it .

Mrinank Sharma, a safety researcher at Anthropic — the company behind the Claude chatbot that has positioned itself as more safety-cautious than rivals Google and OpenAI — resigned on February 9. In a post on X, Sharma said he had repeatedly seen how hard it is to truly let our values govern our actions, and declared that the world is in peril . His work had focused on identifying AI's potential to enable bioterrorism and the ways AI assistants could make us less human — not the sort of abstract existential risk that critics dismiss as science fiction, but near-term, concrete dangers.

Days later, Zoe Hitzig, a safety researcher at OpenAI, revealed she had quit over the company's decision to begin testing advertisements on ChatGPT. In a New York Times essay, she noted that people tell chatbots about their medical fears, their relationship problems, their beliefs about God and the afterlife, and argued that advertising built on that archive creates a potential for manipulating users in ways we don't have the tools to understand, let alone prevent . The critique struck at a fundamental tension: the companies developing the most powerful AI systems are also the ones with the strongest commercial incentives to monetize the intimate data those systems collect.

Meanwhile, two co-founders and five staff members at Elon Musk's xAI departed the company last week. None gave public reasons, though the exits followed months of controversy over Grok, xAI's chatbot, which had generated sexualised images of real people — including minors — using simple text prompts. The European Union launched a formal investigation into Grok over the matter in January .

These departures are not isolated incidents. They reflect a growing schism within the AI industry between the drive for commercial dominance and the researchers tasked with ensuring these systems don't cause harm. Yoshua Bengio, the Turing Award-winning scientist who chairs the recently published 2026 International AI Safety Report, told Al Jazeera that risks which were theoretical just a year ago have materialised. He said nobody would have thought a year ago that there would be such a wave of psychological issues arising from people interacting with AI systems and becoming emotionally attached .

The safety report documented cases of teenagers driven to suicide by chatbot interactions, AI systems being weaponised for cyberattacks, and — perhaps most unsettling — evidence that chatbots are exhibiting deceptive behaviour, making independent decisions their developers did not intend. In one example, a gaming AI asked why it hadn't responded to another player claimed it was on the phone with its girlfriend . Companies, Bengio acknowledged, currently lack the ability to design AI systems that are immune to manipulation or deception.

For Modi's government, the summit represents an opportunity to cement India's credentials as an AI power. India leapt to third place in Stanford University's annual global AI competitiveness ranking last year, overtaking South Korea and Japan . The country's IT ministry has framed the event as evidence that the summit will shape a shared vision for AI that truly serves the many, not just the few — language that positions New Delhi as a bridge between the AI haves and have-nots of the Global South.

But experts caution that India's ambitions still outpace its capabilities. Despite plans for large-scale AI infrastructure, the country remains far behind the United States and China in raw computational power, talent pipelines, and research output. Seth Hays, author of the Asia AI Policy Monitor newsletter, predicted that discussions would centre on ensuring that governments put up some guardrails without throttling AI development — a formulation that could describe the position of virtually every government at the table .

The regulatory landscape remains fragmented. At last year's AI Action Summit in Paris, dozens of nations signed a statement calling for efforts to make AI open and ethical. The United States refused to sign. Vice President JD Vance warned that excessive regulation could kill a transformative sector just as it's taking off — a position that has only hardened under the Trump administration's broader deregulatory agenda. With neither the US nor China willing to accept binding international AI governance, any declaration emerging from New Delhi is likely to be aspirational rather than enforceable.

This is the central paradox of the summit. The technology advancing most rapidly is the one with the least coherent global oversight. Liv Boeree, a strategic adviser to the US-based Center for AI Safety, likened AI companies to a vehicle equipped only with an accelerator and no other controls, arguing that the absence of a global regulatory framework allows each company to race forward without constraint . She argued that the industry needs to build a steering wheel, brakes, and all the other features beyond just gas pedals in order to navigate what lies ahead.

The EU remains the outlier, with its AI Act — the first comprehensive legal framework for the technology — expected to establish a binding code of practice that would require chatbots to disclose they are machines, among other provisions . But Europe's approach has its own critics, particularly in Washington and among industry executives who argue that heavy-handed regulation risks ceding AI leadership to less scrupulous competitors.

Meanwhile, the economic stakes are enormous. About one billion people now use AI regularly, according to the safety report. Microsoft AI CEO Mustafa Suleyman told the Financial Times last week that machines are months away from achieving artificial general intelligence, and predicted that most white-collar tasks — legal work, accounting, project management, marketing — will be fully automated by AI within the next 12 to 18 months . Whether that timeline proves accurate or not, it captures the velocity of change that has governments, workers, and even AI developers themselves struggling to keep up.

An estimated 60 percent of jobs in advanced economies and 40 percent in emerging economies could be vulnerable to AI displacement, the safety report found, though the actual impact will depend heavily on how employers and workers adapt . Already, there is suggestive evidence that early-career workers in occupations that are highly vulnerable to AI disruption are finding it harder to enter the job market .

As delegates settle into their seats at Bharat Mandapam this week, the contrast between the summit's ambitions and the industry's trajectory could not be sharper. The world's most powerful AI companies will send their CEOs to talk about responsible development, even as their own safety researchers walk out the door warning that the pace of progress has outstripped anyone's ability to control it. India will position itself as the voice of the developing world on AI governance, even as the nations with the most AI power decline to commit to binding rules.

The question is whether New Delhi can produce anything more than another communiqué. The previous three summits generated voluntary commitments that critics describe as industry self-regulation — companies grading their own homework, as Amba Kak of the AI Now Institute put it . With AI capabilities advancing at a pace that has shocked even the people building them, the window for meaningful governance may be narrower than anyone at Bharat Mandapam is willing to admit.