Gate News message, April 16 — House Republicans have proposed the Deterring American AI Model Theft Act, which would direct the government to identify entities in China and Russia accused of using improper query-and-copy methods to extract results from U.S. AI models to build rival systems. The bill is set for review by the House Foreign Affairs Committee next week and could prompt Commerce Department blacklisting or presidential sanctions under a 1977 law.
Potential targets include Chinese AI labs DeepSeek, Moonshot, and MiniMax. OpenAI has alleged DeepSeek used ChatGPT results to create a knock-off model, while Google and Anthropic have published reports citing similar unauthorized techniques. A separate House China committee report also urges the Commerce Department to review those firms for entity list action and recommends Justice Department referrals.
The measure is part of a broader legislative push on China, including the Chip Security Act and the bipartisan Guaranteeing Access and Innovation for National AI (GAIN AI) Act. Analysts note such controls could accelerate Chinese technological self-reliance, as occurred after earlier U.S. export controls on Huawei, though the U.S. AI chip advantage remains substantial—currently about five times more powerful, with projections suggesting a 17-fold gap by 2027.