According to a report by technology media outlet CNBC the day before yesterday, the chairmen of two U.S. House committees are conducting a joint investigation and have asked DoorDash, the largest food delivery platform in the United States, to provide information about the Chinese AI foundation models it uses.
DoorDash founder Andy Fang previously posted on X that, according to internal test results, the combination of Moonshot AI’s Kimi K2.6 and Anthropic Fable 5 could outperform Sonnet 4.6 + Opus 4.8 while costing less. The company is currently using a model routing service to assign underlying tasks to Kimi K2.6.
The investigation letter from the U.S. House of Representatives stated: “We recognize that U.S. companies may deploy open-weight models developed in China, which offer more competitive usage costs and customization capabilities. However, these practical considerations do not eliminate security risks or reduce national security concerns.”
Artificial intelligence has become an important area of competition between China and the United States, with both countries vying for a leading position. Andrew Garbarino, chairman of the U.S. House Committee on Homeland Security, previously said: “China is no longer merely following the United States’ development path; it is working to close the gap. Recent reports that Chinese open-source models can reach the level of advanced U.S. models in cybersecurity capabilities are concerning.”
A DoorDash spokesperson responded: “DoorDash firmly supports U.S. AI leadership and is committed to making AI accessible to American businesses. We look forward to working with the investigative committees to discuss how we use AI safely, including cutting-edge models developed in the United States as well as open-weight models.”