By Sergio Boccadutri

On the G20’s agenda there is a dedicated discussion of artificial intelligence and emerging technologies. Right now, innovation and trade ministers are gathering in North Carolina’s Research Triangle for the Innovation Ministerial of the US presidency, on the road to December’s summit in Miami. The draft issue note sets the boundaries of the debate: technological standards, secure supply chains, scientific progress, workers’ skills. All of it within a single frame — innovation as the engine of growth.Here lies the problem, and it is no minor detail. To treat AI as merely one chapter of the economic agenda is to mistake it for just another technology. That is why, at the G20, it cannot remain “one issue among many” on the innovation table: it needs a space of its own.

That this is no ordinary innovation is plain for all to see. Businesses use it to produce and sell better, but so do public administrations, to run healthcare, justice, welfare, borders. When an algorithm assigns a teacher to a posting hundreds of kilometers from home, or decides which taxpayer to single out for an audit, it shapes a public decision that falls on a person. This is a matter of governance, not competitiveness, and it calls for different tools.

Then there is a concrete reason. National rules are diverging: the European AI Act, American executive orders, the laws of individual countries form a mosaic that protects no one. A dedicated space is the natural place to try to bring these approaches into alignment.

Perhaps this is by design. Confining AI to the economy keeps everyone’s hands free: each country stays master of its own rules and free to outrun the others. Yet this is exactly where it goes wrong. Unlike other technologies, nothing guarantees that one state’s advantage in AI will translate into an advantage for its citizens: AI knows no borders, and the risks of failing to coordinate outweigh whatever gains any single country manages to wrest for itself. And there is more. The computing power and the models on which AI rests are by now concentrated in a handful of large companies, a new concentration of power that recalls the one which, more than a century ago, drove the United States itself to build the first antitrust laws against monopolies. Back then it was oil and railroads; today it is data and models. Keeping a public conversation on AI alive (one not confined to a few hands) serves even those states that today believe they can go it alone.

This is why we need a dedicated AI working group, one not subordinate to the economy. Not to slow development down, but to give the governance of these technologies a stable home, one that keeps pace with how they evolve. The G20 already has the power to bring everyone around a table: what is needed is the political will to open a new one — so that the decisions taken in Miami this December, the ones that will shape the coming decade, are not partial on a technology that is anything but partial.