The chairs of two House panels are pressing the food delivery company to explain its deployment of Kimi K2.6, made by Beijing-based Moonshot AI
DoorDash received a letter this week from two House committee chairmen requesting information and documents about its use of artificial intelligence models developed in China, as part of an ongoing congressional investigation into the national security implications of American companies adopting Chinese AI systems.
House Committee on Homeland Security Chairman Andrew Garbarino and House Select Committee on China Chairman John Moolenaar sent the letter to DoorDash, citing a post on X $TWTR by DoorDash founder Andy Fang that described the company delegating lower-level AI work to Kimi K2.6, a model developed by Beijing-based Moonshot AI, according to CNBC.
DoorDash's AI research lab had posted on X that both Kimi K2.6 and Anthropic's Fable 5 beat the Anthropic models it had relied on previously, while also costing less, according to CNBC.
A DoorDash spokesperson said the company "looks forward to engaging with the Committees on how we safely and responsibly use AI, including American-developed frontier models and open-weight models."
The joint investigation began in April with letters to Anysphere and Airbnb $ABNB. The committees said they have since held discussions with both companies about the technical, commercial, and security considerations surrounding their use of Chinese-developed models.
In their letter to DoorDash, the two chairmen acknowledged that cost and capability are driving adoption. "U.S. companies, from large technology firms to startups, may evaluate and deploy PRC-developed open-weight models because they can provide competitive capabilities, lower costs, greater customization, and alternatives to reliance on a small number of proprietary model providers," they wrote. "Those practical considerations do not eliminate the need for risk-based safeguards or diminish the national security concerns associated with growing dependence on models developed by entities subject to PRC jurisdiction."
The committees are examining the origin and training data of Chinese AI models, whether their capabilities were built through unauthorized distillation from leading U.S. systems, and how those models are being deployed or distributed inside the United States. The White House Office of Science and Technology Policy said on July 22 that Moonshot may have operated a covert platform to conduct large-scale distillation against American AI models.
The investigation is also probing whether American open-weight AI offerings are robust enough that U.S. companies have a viable domestic option rather than having to weigh costly or limited homegrown models against inexpensive Chinese ones.
The trend spans coding tools, cloud platforms, and enterprise software, reflecting the growing adoption of Chinese open-weight models across the U.S. tech industry. Unlike proprietary systems, open-weight models can be pulled from public repositories, adapted, and hosted on private servers, a characteristic that has allowed Chinese-origin models to flow into U.S. enterprise environments, including through major American cloud providers like Amazon $AMZN, Microsoft $MSFT, and Google $GOOGL. While some government agencies have banned Chinese AI models such as DeepSeek, U.S. companies are not prohibited from using them. Moonshot unveiled Kimi K3 earlier this month, asserting that the new model has brought Chinese AI capabilities much closer to those of top American systems.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.
