G20 Unanimously Adopts "Carolina Principles," Light-Touch AI Regulatory Framework Goes Global
nashnova research
All G20 members on September 3 unanimously backed the U.S.-led 'Carolina Principles' for AI governance, explicitly rejecting new standalone regulators for AI; this marks the first time Trump's light-touch approach has secured collective endorsement from the world's major economies.
What do the Carolina Principles actually require?
The core demand: no new regulatory bodies for AI. Existing sector regulators — finance, health, transport — each handle AI within their own domain. This means → there will be no single "AI agency"; oversight stays with the authorities that already know the industries.
The framework requires policymakers to factor in the cost of delayed adoption when assessing new technology. In plain terms = regulators must weigh not just the risk of letting AI through, but also the price of holding it back.
It also encourages governments and companies to jointly test emerging technologies, rather than governments writing rules in isolation for companies to follow.
How does this differ from Europe's approach?
The Carolina Principles stand in direct contrast to the EU AI Act: Europe chose new legislation, a new agency, and a risk-tier approval system; the U.S. argues for governing new technology with existing frameworks.
Nvidia CEO Jensen Huang, OpenAI CEO Sam Altman, and Anthropic co-founder Tom Brown all urged delegates not to let safety reviews slow AI model releases. This reflects a collective pushback from Silicon Valley's biggest players against the EU-style regulatory path.
In plain terms = the split comes down to one question: release first and regulate later, or regulate first and release later. The G20 chose the former.
What hurdles remain?
The principles are currently a ministerial-level consensus only. They must still go to the G20 leaders' summit in December — hosted by President Trump at his Doral golf club in Florida — for formal adoption.
This means → between "ministers nod" and "leaders sign," domestic politics in each country can still shift positions, particularly among EU member states.
Whether this light-touch consensus converts into a binding commitment will be the key marker for where global AI regulation is headed.
市场有风险，内容仅供研究参考，不构成投资建议。