At the UN's first Global Dialogue on AI Governance in Geneva, Guterres said the choice is 'govern by design or drift by default' and pushed for harmonised rules to keep AI under human control.
UN Secretary-General Antonio Guterres opened the first Global Dialogue on AI Governance in Geneva on Monday with a warning that global governance of artificial intelligence is falling so far behind the technology's deployment that humanity may already be losing the ability to shape its own future. He gave that loss of agency a name — vibe-coding — a term he used to describe AI systems that write their own code and directly operate machines, networks, and other infrastructure without meaningful human oversight.
The risk, as Guterres described it at the Geneva dialogue, is not just that AI makes bad calls. The human layer is being designed out of the loop, and the systems are no longer waiting for permission to act.
Guterres framed the Geneva meeting as a choice between "governing by design or drifting by default," a binary that, in his telling, is closing faster than policymakers realise. "AI is already transforming our world," he said. "The question is whether we will shape this transformation together, or let it shape us." The warning that AI is moving faster than rules can keep up was carried independently by US News, RTÉ, and Channel News Asia on the same day.
To anchor the structural argument in something concrete, Guterres reached for children. "No child should be a guinea pig for unregulated AI," he said, tying the demand for binding global rules to the protection of children from unregulated AI. The line drew the loudest applause of the morning, but Guterres quickly moved back to the governance pitch: that the alternative to harmonised rules is a patchwork of national regimes none of them can enforce against systems that operate across borders by default.
Nothing binding will emerge from this week's Geneva dialogue. The meeting is consultative, not regulatory, though it is the highest-profile single moment in the UN's nascent AI governance architecture to date, sitting alongside the Independent International Scientific Panel on AI process the body has been building toward. A broader scientific assessment is due next year, and a second dialogue is planned for New York.
Guterres used the moment to land a warning about concentration of power. The handful of companies and countries with the compute and capital to train frontier models, he argued, are making decisions for everyone else. The choice to convene in Geneva is itself a signal that the UN wants to anchor the rules conversation outside any single national capital.
A UN press release on the opening day carried the same remarks and confirmed the week's programme. Outside the venue, civil society groups pressed for binding commitments on transparency, accountability, and the use of AI in public services.
The next checkpoint is the scientific panel's full assessment, due in 2027. The Geneva dialogue puts deliberate, harmonised governance on one side and continued drift on the other.