Representative Ro Khanna, the ranking Democrat on the House Select Committee on the Strategic Competition Between the United States and the Chinese Communist Party, has initiated a formal inquiry into AI security. On October 1, 2026, he sent letters to the CEOs of OpenAI, Anthropic, Google, Meta, and SpaceXAI requesting reports on any known attempts by China or other foreign actors to steal their model weights.
The inquiry seeks detailed information regarding illegal access attempts and the current cybersecurity measures these companies employ to protect their intellectual property. Khanna emphasized that the theft of model weights could compromise American leadership in the artificial intelligence sector.
Beyond direct theft, the congressman expressed concern over model distillation, where developers train new systems using outputs from existing US models. He specifically mentioned Chinese firms Moonshot AI and DeepSeek as entities that may be exploiting this technology.
Khanna has previously addressed these risks, including holding a hearing in April 2026 and communicating directly with Chinese labs such as DeepSeek, Alibaba, and Moonshot AI. He has also consulted with US intelligence officials regarding the national security implications of AI technology.
