The US will ask the G20 not to build anything to govern AI

The United States will ask the G20 not to build any new institutions for governing artificial intelligence. A White House official set out the position before a two-day meeting that opened in North Carolina on Tuesday, Reuters reported.

The argument is procedural rather than substantive, which is what makes it effective. Washington is not proposing rules it prefers; it is pressing member countries not to establish bodies that could write rules at all.

Commerce Secretary Howard Lutnick and White House technology adviser Michael Kratsios are hosting.

The session is the first in a series running to the G20 leaders’ summit in Miami in December, which means the position set out this week is the one other countries will spend three months negotiating against.

The guest list tells you how the administration sees the subject. Sam Altman and Jensen Huang are attending in person and Elon Musk by video, which is an unusual composition for an intergovernmental meeting.

Canada has signalled it will push the other way, arguing for a balance between innovation and public trust and safety. China’s embassy did not say whether it would attend.

The G20 has no power to compel anybody, which cuts both ways. It cannot create a regulator, and the language it agrees still becomes the reference point national governments cite when they act at home.

The administration made the same case at the Paris AI summit, where JD Vance warned against paralysing the technology with regulation, and the US declined to sign an international statement on regulated AI use.

What has changed is the forum and the stakes. A G20 process produces communiqué language that shapes what national regulators believe they have licence to do, and an agreed sentence about international oversight is harder to unpick than a summit speech.

Europe arrives at this meeting having already built what the US is arguing against. The AI Act is in force, the Digital Services Act is being enforced against ChatGPT and other designated platforms, and the Cyber Resilience Act took effect this month.

The timing is awkward in another way. The chair of the Financial Stability Board told G20 finance ministers this week that AI-driven cyber risk is the most immediate threat to financial stability, and that warning came through exactly the kind of multilateral body the US wants left alone.

A United Nations panel has separately warned that AI development is outpacing both scientific understanding and government policy. That is the case for new institutions, made by an existing one.

For European regulators, the practical question is narrower than the principle. Nothing agreed in Miami would undo the AI Act, but a G20 consensus against international oversight makes European enforcement look like an outlier rather than a template.

The counter-argument deserves stating properly. Institutions built now would be designed around a technology nobody understands well, staffed by people appointed for a landscape that will have changed, and slow to correct once established.

There is also a competitive dimension nobody at the meeting will state plainly. A country hosting most of the leading model developers has less to gain from international oversight than countries that host none.

Developing economies have a different interest again, and a quieter one. Countries that neither build models nor regulate them effectively have most to gain from international standards and least capacity to insist on them.

Whether the position holds depends on who else wants a fight. The G20 works by consensus, and a communiqué that says nothing about AI governance is a win for the argument that nothing needs governing.

Miami is in December. Between now and then, the question is whether any member is willing to spend the diplomatic capital required to write a sentence the United States has said it will not accept.