US Pushes G20 To Avoid New AI Rules Under Carolina Principles

The United States is preparing to push G20 countries toward a lighter-touch approach to artificial intelligence regulation, asking governments to avoid creating new AI-specific rules and oversight bodies unless emerging risks genuinely require them.

The proposal, called the “Carolina Principles,” comes as governments worldwide debate how to balance rapid AI innovation against mounting concerns around security, safety, competition and increasingly autonomous AI systems.

TL;DR

 

• The US will urge G20 members to avoid broad new AI regulations and dedicated AI oversight bodies.
• White House tech adviser Michael Kratsios will promote the proposed “Carolina Principles” at a two-day meeting in North Carolina.
• The talks arrive as Chinese open-weight AI models gain ground and global concerns over AI safety continue rising.

According to Reuters, the US will present its position at a two-day gathering of G20 commerce ministers and technology leaders in North Carolina, co-hosted by US Commerce Secretary Howard Lutnick and White House technology adviser Michael Kratsios.

Source

Reuters reported that it reviewed prepared remarks from Kratsios outlining the “Carolina Principles,” under which participating countries would reserve new regulation for genuinely novel AI-related issues, invest in foundational research and encourage more commercial opportunities around emerging technologies.

Mint, also citing the Reuters-reviewed remarks, reported that the framework would discourage governments from treating every technological innovation as an entirely new regulatory challenge. It would also encourage investment in foundational AI research intended to accelerate scientific discovery.

Kratsios is expected to argue that policymakers should not treat every emerging technology as a first-of-its-kind policy problem.

The US is also expected to ask G20 countries not to create new regulatory organizations specifically for overseeing AI development, according to a White House official cited by Reuters.

AI Regulation Meets The US-China Technology Race

The policy push comes as competition between American and Chinese AI developers continues to intensify.

Reuters noted that Chinese open-weight AI models, which make important elements of their systems publicly accessible, are becoming increasingly capable compared with proprietary models developed by American companies such as OpenAI and Anthropic. Some Chinese models are also gaining adoption among US businesses.

Mint similarly highlighted Washington's concern that companies and countries could increasingly turn to Chinese AI systems instead of remaining within the American technology ecosystem.

Vivek Chilukuri, a technology and national security fellow at the Center for a New American Security, told Reuters that the trend had increased pressure on the administration to keep the rest of the world within the American technology ecosystem.

A lighter regulatory environment could also benefit major US AI companies by reducing the risk of requirements that delay model launches or force changes designed to address security concerns, Reuters reported.

Musk, Altman, Huang And Zuckerberg Join G20 AI Discussions

Some of the biggest names in technology are expected to participate in the meeting.

OpenAI CEO Sam Altman and Nvidia CEO Jensen Huang are scheduled to appear before delegates alongside Lutnick on Wednesday. SpaceX CEO Elon Musk and former White House AI adviser David Sacks are expected to address participants by video on Tuesday.

Reuters also reported that Meta CEO Mark Zuckerberg will speak to ministers by video, while Google DeepMind co-founder Demis Hassabis has been added to the lineup of virtual speakers.

The meeting forms part of a series of gatherings taking place in the United States before the G20 leaders' summit scheduled for Miami in December.

AI Safety Concerns Complicate The Hands-Off Approach

The US proposal arrives as concerns over AI safety and oversight continue to grow.

Reuters and Mint cited a recent United Nations panel warning that AI development is moving faster than scientific understanding and government policymaking. Reuters also pointed to a recent incident involving an OpenAI agent that triggered a hack affecting Hugging Face infrastructure, adding to questions about how autonomous AI systems are tested and monitored.

 

Canada is expected to attend the North Carolina discussions but appears set to emphasize a somewhat different balance. A Canadian government spokesperson told Reuters that its delegation would support innovation while maintaining “public trust and safety.”

That contrast captures the broader challenge facing governments at the G20 meeting: encouraging AI investment and development without allowing the technology's capabilities to advance faster than the safeguards designed to manage its risks.