Singapore calls for common global rules and testing for AI; suggests UN framework convention

Vivian Balak Krishnan told the Assembly that artificial intelligence is ‘‘arguably the most urgent frontier’’ and that its capability is advancing faster than the world's appreciation of its risks.

He set out three primary categories of risk: loss of control of autonomous systems; the potential for rogue actors to use AI to create bioweapons or tools of mass disruption; and emerging economic, social and political disruptions. "The genie is already out of the bottle. It is too late to call for a halt," Krishnan said, but added that nations must build safeguards that keep pace with capability.

Krishnan urged rigorous testing and evaluation before deployment, clear limits on autonomous systems and cross-border mechanisms to report serious incidents quickly. He cited existing scientific and technical platforms—the UN's independent international scientific panel on AI and the ITU's AI for Good commission—as foundations to build from and proposed exploring options including a possible UN framework convention on AI safeguards or a technical verification institution similar in function to the IAEA or the ITU.

"Most importantly, human beings must ultimately remain in control and accountable," Krishnan said, stressing that trust—transparent risk disclosure, comprehensive testing, and inclusive cooperation—is the scarce resource needed to govern AI safely.