Congress Opens Inquiry Into Chinese AI and Data Security Concerns

Two top Republicans on the House Homeland Security Committee are launching an investigation into artificial intelligence (AI) models developed by the People’s Republic of China (PRC).  

Rep. Andrew Garbarino, R-N.Y., and Jim Moolenaar, R-Mich., who serve as the chair of the House Homeland Security Committee and Select Committee on China, respectively, announced the probe on Wednesday.  

The investigation follows allegations made last week by the White House that the PRC was running industrial-scale distillation campaigns to steal U.S. AI secrets and technologies. AI distillation is a technique in which actors query proprietary models tens of thousands of times to mimic their behavior.  

Distillation can help create cheaper and more efficient versions of more advanced and larger AI models. Some of the models Garbarino and Moolenaar said they will investigate include DeepSeek, Alibaba, Moonshot AI, and MiniMax, which are all developed by Chinese companies.  

In their announcement, the congressmen said that AIs that are the result of unauthorized distillation could be “marketed or made available to U.S. companies, developers, and consumers,” while not including safeguards embedded in the original American model.  

Two letters were sent to U.S. companies, Anysphere – the parent company of Cursor – and Airbnb, as an initial part of the lawmakers’ investigation. Garbarino and Moolenaar voiced concerns about the companies’ use or exposure to PRC-made AI models. The congressmen noted that the AI model used by Anysphere was reportedly built on an open-weight model developed by Moonshot AI, and Airbnb used an Alibaba model for customer service operations.  

“Beijing has been explicit in its view that the global distribution of open-weight AI serves PRC strategic interests,” the lawmakers wrote to Anysphere. “What is at issue, therefore, is not simply market competition, but the growing risk that software systems used across the American economy, government, and defense industrial base will come to depend on models developed by PRC-linked laboratories and shaped by PRC strategic objectives.”   

To Airbnb, which reportedly said it was relying on Alibaba’s Qwen model over U.S. alternatives to save costs and boost speed, the congressmen wrote, “The Committees have serious concerns about the national security and data-security implications of that approach for Airbnb’s American customers and for the integrity of its systems.”   

“[T]he spread of Chinese open-weight AI models carries consequences well beyond ordinary software adoption preferences… American firms adopting these models are not simply choosing a cheaper tool, they are importing an architecture designed to serve the Chinese state,” the letter continued. 

The lawmakers did not disclose what the next steps may be in their investigation into PRC-aligned distillation campaigns and their impacts on U.S. national security. 

PRC AI companies have used distillation efforts to train from leading U.S. models, including Anthropic. The company said in February that it identified industrial-scale campaigns launched by three China-affiliated AI laboratories – including DeepSeek – which were attempting to “illicitly extract Claude’s capabilities to improve their own models.”  

According to Anthropic, the three firms generated over 16 million exchanges with Claude and created approximately 24,000 fraudulent accounts.