Trump administration’s internal divisions shape US AI policy amid China competition

The Trump administration is relying on a small group of officials across multiple agencies to shape US AI policy amid rising competition with China, but internal divisions are complicating decision-making. One senior White House official described the situation as “an argument with ten sides,” reflecting the lack of interagency coordination and the competing priorities among policymakers.

Commerce Secretary Howard Lutnick has emerged as a central figure in these debates, overseeing export controls through the Bureau of Industry and Security. Lutnick has explored incentives to encourage US labs to develop open-weight models as a counterbalance to Chinese advancements, while also imposing export controls on Anthropic. He has publicly questioned the capabilities of Chinese models, such as Moonshot AI’s Kimi K3, stating that US frontier models outperform them in key benchmarks.

Arvind Raman, acting director of the Center for AI Standards and Innovation (CAISI) within the Commerce Department, has taken a more collaborative approach with industry, including working with Anthropic to strengthen safeguards for its Fable 5 model. This followed weeks of negotiations to address concerns about model jailbreaks.

National Cyber Director Sean Cairncross has adopted a harder line toward Chinese AI labs, focusing on national security risks. Cairncross played a key role in crafting the June 2 executive order that established a framework for assessing the most powerful AI models. Despite lacking a technical background, his staff has been empowered to address complex AI challenges without excessive White House micromanagement.

White House Chief of Staff Susie Wiles serves as the final arbiter for policy proposals, vetting and scrutinizing them before they reach President Trump. Her decisions are described as the decisive factor in shaping administration policy on AI.

Treasury Secretary Scott Bessent has taken an aggressive stance toward Chinese AI, calling model distillation—where Chinese labs train models using US models—“IP theft” and threatening sanctions or inclusion on the US government’s Entity List. Bessent has urged Western allies, including Europe and Japan, to accelerate development of open-weight models to avoid falling behind China.

Former AI Czar David Sacks, though no longer in an official role, remains influential, advocating for a hands-off approach to AI regulation and publicly supporting US leadership in frontier models. His public statements, including on social media, have amplified his influence within the administration.

The Treasury Department has also focused on risks posed by AI in financial systems, citing potential scenarios where rogue models could disrupt markets or access sensitive banking information.