All 20 G20 Nations Endorse US-Proposed Carolina Principles AI Framework

All 20 G20 member nations, including China and Russia, unanimously endorsed the U.S.-proposed Carolina Principles at a two-day summit. The framework encourages sector-specific AI regulation, discourages establishing new regulatory agencies, and promotes close collaboration with private industry to evaluate emerging technologies. White House science and technology advisor Michael Kratsios promoted the principles at the summit, and a central commitment is that countries should 'reserve new regulation for novel considerations,' discouraging treating every new AI capability as requiring a new regulatory rulebook. The accord is non-binding but is expected to be formally adopted at the G20 leaders summit in December at Trump National Doral in Florida, weeks before a planned Trump-Xi meeting in Washington where AI governance is a key agenda item.

OpenAI CEO Sam Altman appeared at the summit alongside U.S. Commerce Secretary Howard Lutnick, comparing a country rejecting AI to a country rejecting electricity and arguing AI is necessary for governments to deliver the quality of life citizens should expect. Altman also stated that work previously taking startups three months can now be completed in approximately 17 minutes with AI. Nvidia CEO Jensen Huang also spoke, describing AI investment as reaching trillions of dollars and arguing AI could add $20 to $50 trillion in global economic benefit; Huang also helped apply pressure in support of the Carolina Principles. The U.S. delegation experienced internal tensions between the Commerce Department and OSTP during the summit.

Following the endorsement, Beijing signaled through a CCTV-affiliated account that U.S. AI companies such as Anthropic should be subject to the same frontier AI safety restrictions Washington seeks to impose abroad, positioning Anthropic as a test case for challenging U.S.-defined AI governance. China's position, per the signal, is that dangerous AI capabilities should be jointly defined and controlled rather than unilaterally by the U.S., while ordinary Chinese AI competition should remain outside frontier safety restrictions. The signal came through an unofficial channel rather than a formal government announcement.