US Urges G20 to Adopt Light-Touch 'Carolina Principles' on AI

The White House is pressing fellow G20 governments to adopt a light-touch framework for regulating artificial intelligence, ahead of a two-day digital economy meeting opening September 3 in Chapel Hill, North Carolina, according to AFP reporting on the plan.

What the Carolina Principles propose

The framework, named the Carolina Principles after the meeting’s host state, asks governments to rely on existing laws wherever they already cover AI and to write new AI-specific rules only for genuinely “novel” problems current law doesn’t address. It also calls on countries to steer public funding toward foundational AI research and to avoid setting up new regulatory bodies dedicated solely to AI.

White House Office of Science and Technology Policy Director Michael Kratsios and Commerce Secretary Howard Lutnick are organizing the meeting at the University of North Carolina, part of a broader push under the United States’ rotating G20 presidency this year. A separate G20 leaders’ summit is scheduled for mid-December in Miami.

As the meeting opens, no country has formally signed on to the principles — the US is presenting them as a proposal rather than an agreement already in place. Canada, among the expected attendees, has struck a more cautious tone, with officials saying any approach should preserve “public trust and safety” alongside innovation.

A contrast with the EU

The pitch targets governments moving toward stricter oversight, chief among them the European Union, whose AI Act imposes binding rules on high-risk systems. Speaking ahead of the meeting, Tesla and xAI’s Elon Musk, participating virtually, argued that “the regulation level is extraordinarily high” in the EU and said new technologies “must be default legal as opposed to default illegal.” EU digital chief Henna Virkkunen is expected to attend, along with officials from France, Italy and Germany.

Executives including Nvidia’s Jensen Huang, OpenAI’s Sam Altman, Google DeepMind’s Demis Hassabis and Meta’s Mark Zuckerberg are also expected at the gathering, a sign of how directly the industry’s largest players are shaping the regulatory debate the US is trying to steer.

The proposal follows a string of unresolved US AI-oversight questions at home, including the White House’s own frontier-model review framework, which it has kept confidential even as it presses other governments toward lighter rules abroad.