Chinese AI models and US federal policy

House Homeland Security Chairman Andrew Garbarino and Select Committee on China Chairman John Moolenaar subpoenaed DoorDash on August 9 to explain its use of Kimi K2.6, a model from Beijing-based Moonshot AI, after co-founder Andy Fang posted that the company routes lower-level AI work to it.

DoorDash's own AI research found Kimi K2.6 and Anthropic's Fable 5 outperformed its previous models at lower cost. The same panels sent letters to Cursor, Airbnb, and Siemens over comparable Chinese AI use, and the FRONTIER Act was introduced to give congressional AI oversight a legislative form.

BIS reported investigating suspected smuggling routes and remote access pathways for Chinese firms to access restricted Nvidia chips

On August 7, the White House paused planned sanctions and cloud access bans on Chinese open-source AI models after Nvidia, Meta, Microsoft, and Google lobbied against them, while chairs of two House Select Committees wrote to DoorDash over national security concerns about its use of Kimi K2.6. Representative Lori Trahan cited two AI containment breaches within one month and called for the FRONTIER Act. Her Great American AI Act of 2026, co-authored with Representative Jay Obernolte, would nationalize frontier-model governance, and a Trump executive order requiring pre-release government model sharing added a third federal track.

Representatives Trahan and Casar called for oversight hearings and legislation following reported AI containment breaches; House committee chairs had raised national security concerns about DoorDash's Kimi K2.6 use

White House paused planned sanctions and cloud access bans on Chinese open-source AI models after Nvidia, Meta, Microsoft, and Google lobbied against them