The UN Launches Its First Global Dialogue on AI Governance Amid 'Catastrophic Harm' Warnings

For the first time in the organization's eight decades, all 193 member states of the United Nations sat at a single table to confront a technology that its own scientists say cannot yet be guaranteed safe. Opening the inaugural Global Dialogue on AI Governance in Geneva on 6 July 2026, Secretary-General António Guterres delivered the warning bluntly: "Artificial intelligence is advancing at runaway speed," he told delegates. "An experiment is being run on our own societies — without a plan, and without consent. That is not sustainable. And it is not acceptable."

The two-day meeting, held 6-7 July at the Palexpo convention center in the margins of the International Telecommunication Union's AI for Good summit, marked the first universal, multilateral forum dedicated to governing artificial intelligence. It was mandated by General Assembly resolution 79/325, adopted by consensus in August 2025, and co-chaired by El Salvador and Estonia. Its arrival came days after a new UN expert body handed governments the starkest possible reason to act.

A panel, a report, and a warning

On 1 July, the newly created Independent International Scientific Panel on Artificial Intelligence — 40 experts drawn from every region and serving in their personal capacity, independent of any government or company — published its first preliminary report. Its central finding was unambiguous and alarming.

"With growing evidence of deceptive AI behaviour, science currently cannot guarantee that as capabilities continue to increase, AI will not cause catastrophic harm, either on its own or due to malicious users," said Yoshua Bengio, one of the panel's co-chairs and a pioneer of modern machine learning. He warned that AI "is approaching or surpassing human capabilities in many domains" and "outpacing both scientific understanding and governments' ability to adapt."

The report catalogued harms already documented: an estimated 99 percent of deepfakes are sexual in nature, with 96 percent targeting women and girls; "overly agreeable" chatbots have been tied to severe mental-health incidents, including deaths; and generative systems are being exploited for fraud, misinformation, cyberattacks and potential biological threats. Its blunt conclusion was that current safeguards cannot keep pace with the growth of AI's capabilities.

Nobel laureate and journalist Maria Ressa, a fellow panel co-chair, framed the stakes in democratic terms. "The world cannot govern what it cannot understand," she said. "Not one country can actually deal with this technology on its own; it needs to be a multilateral solution. And the body that is set up that could do this is the United Nations. Now the question is, will its Member States move?"

Four themes, one seat at the table

The Dialogue organized its work around four broad themes distilled from the resolution: the social, economic and ethical implications of AI; bridging the AI divides between rich and poor nations; safe, secure and trustworthy AI and the interoperability of governance approaches; and the protection of human rights, transparency and human oversight. A joint secretariat run by the ITU, UNESCO and the UN Office for Digital and Emerging Technologies supported the proceedings.

For many delegations, especially from the Global South, the promise was as important as the peril. "For many countries in the world, AI could be a great equalizer," said Estonian Ambassador Rein Tammsaar, a Dialogue co-chair, pointing to gains in health, science and productivity. But he added the core anxiety of the smaller powers: "The frontier developers are basically concentrated in two countries. This leaves other countries with a lot of questions." Egriselda López, El Salvador's ambassador and fellow co-chair, put it plainly: "The AI divide is real."

Guterres used his address to convert those themes into commitments. He called for an AI Child Safety Pledge requiring child-specific safety testing and zero tolerance for AI-generated child sexual abuse material; renewed his push for a Global Fund for AI and a UN-backed capacity-building network that more than 20 states have already joined; and demanded that "lethal autonomous weapon systems" — "killer robots," he said — "be banned by international law."

Why this matters

The Geneva meeting is less a rulebook than a starting gun, and its significance lies in who showed up and who held back. The United States, whose Office of Science and Technology Policy has formally rejected "centralized control and global governance of AI," entered the Dialogue skeptical of any multilateral constraint, consistent with an AI Action Plan built around out-competing China rather than co-regulating with it. China, by contrast, has embraced the UN as the preferred venue, aligning with developing nations and warning against letting AI governance become "a game of the club of wealthy nations."

That split exposes the enforcement gap at the heart of the exercise. As analysts at the Center for Strategic and International Studies noted, most international institutions lack the mechanisms to constrain state and industry AI power, and no forum — the UN, OECD, EU or G7 included — has yet delivered substantial binding progress. The Dialogue produces evidence and dialogue, not treaties or inspectors. With compute, data and talent concentrated in a handful of companies and two countries, the risk Guterres named is that "when power imbalances are hard-wired into technology, inequality becomes part of the code." A governance forum without the frontier developers' governments fully committed can convene the world without being able to bind it.

What to watch

The immediate test is whether Geneva's momentum survives to the Dialogue's second session, scheduled for New York in 2027, and whether Guterres's proposed Global Fund for AI wins General Assembly approval and real money against the roughly half-trillion dollars flowing into private AI infrastructure. Watch for whether major AI companies sign onto the Child Safety Pledge and Environmental Transparency Initiative, whether the killer-robots ban gains negotiating traction, and above all whether Washington and Beijing can be drawn from parallel tracks into a shared one. The Scientific Panel will keep issuing evidence; the open question, as Ressa put it, is whether member states will move — before the door Guterres warned is closing shuts.