Why a DoorDash Co-Founder's Tweet About Chinese AI Triggered a Congressional Investigation

That economic efficiency caught the attention of two powerful House Republicans who had already been investigating Chinese AI adoption for months.

What Lawmakers Are Demanding

On July 31, 2026, House Select Committee on China Chairman John Moolenaar (R-MI) and House Homeland Security Committee Chairman Andrew Garbarino (R-NY) sent a formal letter to DoorDash co-founder and CEO Tony Xu . The letter requested :

• By August 14, 2026: A detailed briefing and documents on all PRC-developed open-weight AI models DoorDash has evaluated, tested, or deployed, including the specific business purposes for each model and the company's security risk assessment policies.
• By August 21, 2026: A broader production of documents, including internal communications, risk assessments, security evaluations, and any agreements with Chinese AI developers.

The lawmakers explicitly asked DoorDash to identify every Chinese AI model the company has used and to explain its evaluation and risk mitigation processes .

The National Security Concerns: Back Doors and Sleeper Agents

In their letter and press release, Moolenaar and Garbarino warned that Chinese open-weight AI models could contain hidden vulnerabilities. The committees specifically stated that "back doors or other malicious functionalities—sometimes called 'sleeper agents'—could be inserted into models without users' knowledge" .

Garbarino told CNBC: "The Chinese Communist Party is no longer just nipping at our heels in artificial intelligence; it is racing to close the gap in some of the exact capabilities that will shape the future of cybersecurity."

However, the lawmakers have acknowledged there is no widespread public evidence that such back doors or sleeper agents have actually been discovered in production Chinese models being used by US companies. The concerns are framed as preemptive and about potential risk, not confirmed compromise. The investigation letter itself notes that the committees recognize companies may evaluate PRC models for "legitimate technical reasons" but want to understand whether adequate safeguards exist .

A Broader Pattern: The Economy of Open-Weight AI

DoorDash's case is the latest—and most consumer-facing—example of a broader trend. In April 2026, the same two committees opened a joint investigation into Airbnb and Anysphere (maker of Cursor) over their adoption of Chinese-developed AI models . The committees stated that "U.S. companies, from large technology firms to startups, may evaluate and deploy PRC-developed open-weight AI models" due to their lower cost and competitive performance .

The DoorDash case is particularly notable because the company's own benchmarking showed the Chinese model was not just cheaper—it was more effective for the specific task of code review, making the economic incentive clear . As one analysis put it, "the winner was not the all-premium configuration."

DoorDash's Response

DoorDash has not issued a detailed public response to the investigation letter as of early August 2026. However, a spokesperson told CNBC: "DoorDash proudly supports American AI leadership and is working to ensure AI benefits Main Street, not just the biggest companies. We look forward to engaging with the Committees on how we safely and responsibly use AI, including American-developed frontier models and open-weight models."

The company's posture, reflected in its published AI research and Fang's posts, is pragmatic and transparent: it uses Chinese open-weight models only for lower-level code review, not for customer-facing systems or sensitive data, while reserving US frontier models (Anthropic's Fable) for the most complex work . DoorDash's AI research team emphasized that DashBench allows it to "quickly discern which model combinations yield the best results and at the best cost."

What Comes Next

The investigation seeks to establish a "factual basis" for potential federal policy on the use of PRC AI models by American firms . The committees' letter concedes that Chinese open-weight models offer competitive capability at lower cost, then argues that "effective federal responses should focus on enhancing the availability, safety, and competitiveness of American open-weight alternatives" while applying tailored safeguards in sensitive applications .

DoorDash is not accused of breaking any law. US companies may freely run Chinese open-weight models today . But with an August 14 briefing deadline and an August 21 document deadline, the company is now at the center of a debate that could reshape how American businesses think about the provenance of their AI models.