Trump Administration Reveals AI Policy Divisions Ahead of G20

The AI policy team of the Trump administration has shown internal friction regarding global AI leadership ahead of the G20 Innovation Ministers' meeting, with the Commerce Department and the White House Office of Science and Technology Policy yet to form a fully consistent action framework for external strategy.

The publicly confirmed common stance is that the U.S. will push G20 members to adopt a "light regulatory" AI policy and oppose the establishment of a new dedicated AI regulatory agency. Michael Kratsios, Director of the White House Office of Science and Technology Policy, and Commerce Secretary Howard Lutnick co-hosted a meeting in North Carolina to encourage countries to accept the "Carolina Principles."

The core of the Carolina Principles includes: investing in fundamental scientific research, expanding commercial opportunities for emerging technologies, creating new regulations only when genuinely novel issues arise, and avoiding the establishment of new AI regulatory bureaucracies. The U.S. hopes to embed a policy orientation of "default allowing innovation, intervening only after specific risks arise" into the international coordination framework.

The focus of the divisions is not necessarily whether to support AI, but rather what tools the U.S. should use to lead global rules: the Commerce Department emphasizes industrial competitiveness, exports, commercial deployment, and collaboration with businesses; the White House science policy team is more responsible for interdepartmental policy coordination, technical standards, international principles, and the overall presidential agenda. If the two systems present different statements regarding commitments to allies, regulatory language, export controls, or standard-setting, external partners may lower their expectations for the continuity of U.S. policy.

The meeting invited tech leaders such as Elon Musk, OpenAI CEO Sam Altman, and Nvidia CEO Jensen Huang, indicating the government's attempt to integrate model companies, chip manufacturers, cloud infrastructure, and international trade policy into the same diplomatic context. The actual needs of businesses are not entirely the same: model companies focus on computing power, electricity, data, and accountability rules; chip companies prioritize export markets and supply chains; cloud companies are concerned with data center licensing, energy, and cross-border data regulations.

The external challenge for the U.S. is that the EU has established a risk-based regulatory system through the AI Act, while China shapes its domestic framework with filing, content, and model governance rules, and the U.S. aims to avoid the establishment of a broad federal AI regulatory agency. If Washington cannot provide a consistent combination among "promoting commercial deployment," "restricting high-risk uses," "exporting advanced chips," and "unifying standards with allies," countries may continue to adopt EU-style compliance frameworks, Chinese-style national governance frameworks, or fragmented domestic rules.

In market mechanisms, buyers rely on global markets, cross-border data, computing power supply, and unified compliance standards, including AI model companies, chip manufacturers, cloud service providers, and enterprise users; sellers or rule providers are national regulatory agencies, standard organizations, cloud infrastructure, and compliance service providers. The U.S. push for low regulation may reduce short-term compliance costs for domestic model and chip companies, but if international rules lack consistency, businesses will face multiple audits, data localization, and market access requirements; beneficiaries will be large platforms and compliance suppliers that can meet U.S., EU, and Asian regulatory requirements simultaneously, while resource-scarce startups will bear higher costs for cross-border expansion.

Source: Public Information

ABAB AI Insight

The AI policy of the Trump administration is not a single document but is composed of the White House Office of Science and Technology Policy, the Commerce Department, the Defense System, the Energy System, and the Trade and Export Control Department. The Commerce Department controls export regulations, chip trade, industrial policy, and business relations, while the White House Office of Science and Technology Policy is responsible for translating presidential strategies into interdepartmental agendas; both can claim to represent "U.S. AI leadership," but during external negotiations, industrial interests, national security, ally coordination, and regulatory principles often do not point to the same answer. The G20 meeting co-hosted by Kratsios and Lutnick reflects that technology policy and business strategy must be handled in tandem.

In terms of capital pathways, the U.S. AI advantage relies on four resources that can be changed by policy: the supply of advanced chips from companies like Nvidia, the data centers and power assets of cloud providers, the funding and training data of model companies, and the revenue and developer ecosystem brought by overseas markets. A low regulatory approach can accelerate domestic deployment and reduce compliance costs for startups; however, implementing export restrictions on advanced chips may limit some overseas revenue and stimulate other countries to establish local alternative supply chains. If the policy team cannot clearly define when to prioritize commercial expansion and when to prioritize security restrictions, businesses will struggle to decide on data center investments, product regional launches, model openness, and long-term supply contracts.

Historically, internet governance has also seen a competition between "U.S. corporate innovation first" and "EU privacy regulation first". The GDPR ultimately became a de facto standard that multinational tech companies must adapt to, as businesses entering the European market need to bear its compliance requirements; the AI Act may follow the same spillover path. The U.S. proposing the Carolina Principles is an attempt to prevent AI rules from being locked into a high regulatory model first: if G20 countries do not establish new AI regulatory agencies, leading U.S. companies can enter markets with lower institutional friction. However, a light regulatory approach without clear security commitments may also lead some allies to turn to the EU framework for stronger consumer and rights protection narratives.

This falls under regulatory change. International competition in AI does not only occur in model rankings but also in "who defines compliance costs, who decides data flow, who can export chips, and who bears responsibility for model accidents" at the rules level. Internal strategic divisions within the government will reduce the efficiency of the U.S. converting domestic industrial advantages into international institutional advantages: businesses need predictable rules, allies need sustained commitments, and regulators need credible enforcement. If these conditions are lacking, the global AI market will form different models, data, cloud, and security standards by region, and leading technologies will also be weakened by fragmented market access rules.

ABAB News · Cognitive Law

Technological leadership does not equal regulatory leadership
Inconsistent domestic policies will outsource international standards
The more fragmented the regulation, the easier it is for giants to win