Singapore's Foreign Minister Vivian Balakrishnan on 27 September 2026 called for formal negotiations on a United Nations convention on artificial intelligence, warning that governments worldwide face serious and underappreciated risks from autonomous systems even as they race to capture the technology's economic benefits. His remarks were delivered at an Asia Society discussion on the sidelines of UN proceedings in New York.

Three Core Risks Flagged

Balakrishnan identified three broad categories of AI risk that he said demanded multilateral attention. The first is the potential loss of human control over AI agents. 'Anyone who tells you, even the creator of the AI, that he's got absolute control of the kill switch is fooling you,' he said, adding that 'loss of control is a very real risk.'

The second concern is the threat from rogue actors: Balakrishnan warned that AI could dramatically lower the barrier for bad actors to develop biological or chemical weapons. Third, he highlighted the corrosive effect on public trust from AI-generated synthetic media — the ease with which convincing images and videos can be fabricated, he argued, leaves people less certain of what they see and hear, eroding the foundations of political debate.

The Case for a UN Framework

Despite his stark warnings, Balakrishnan was explicit that he is not calling for a halt to AI development. 'I'm not a believer in putting a halt to AI development,' he said. 'I'm not even a believer in, you know, just putting the brakes on AI.' Instead, he compared AI to driving a fast car: a powerful engine needs brakes, seat belts, and rules of the road.

He pointed to international agreements on the law of the sea and climate change as models — difficult but ultimately necessary multilateral negotiations. The United Nations, he argued, offers a uniquely inclusive forum where countries of vastly different sizes and technological capabilities can participate as equals, making it the appropriate venue for such a convention.

The US-China AI Dynamic

Balakrishnan acknowledged that the global AI landscape is being shaped primarily by the competition between the United States and China. He estimated that America holds a lead of perhaps six months in frontier AI models, but noted that China is determined to seize the opportunities of the new technological era. He welcomed a reported discussion on AI between US President Donald Trump and Chinese President Xi Jinping, while acknowledging concerns that the guardrails emerging from their talks remained thin. 'You need to start somewhere, and I'm glad that that discussion is going on,' he said. Their bilateral rivalry, he cautioned, would have profound consequences for Southeast Asia.

Southeast Asia's Strategy: Deploy, Don't Compete

For Singapore and its neighbours, Balakrishnan outlined a pragmatic path: gain access to technology, keep supply chains open, and put AI to work across their economies. He conceded that Southeast Asian nations are unlikely to compete with the two powers in building frontier models, but argued the region could excel at deploying AI in public services and other applied fields.

That ambition would require significant changes in education and ongoing workforce retraining as the nature of work evolves. Balakrishnan said Singapore's immediate focus is on workers whose day-to-day tasks will be transformed by AI. Whether any particular job disappears entirely is uncertain, he said, but governments have an obligation to prepare citizens for the disruption ahead.

What Comes Next

Balakrishnan's call adds diplomatic momentum to a growing chorus of voices — from academics to technology executives — urging a binding international framework for AI governance. No formal UN process has yet been launched for such a convention, and significant geopolitical hurdles remain. Whether the Trump-Xi dialogue on AI can serve as a foundation for broader multilateral engagement will be closely watched in the months ahead.