Blocking "Industrial-Scale Theft"... US Government to Counter Foreign AI Model Theft and Strengthen Export Controls
The White House and the Department of Commerce have defined the theft of artificial intelligence (AI) models by foreign powers, including China, as 'industrial-scale' and are launching a new export control framework and the 'American AI Exports Program' to block it.
On Friday, April 24, 2026, the White House Office of Technology Policy warned that foreign powers are engaging in 'industrial-scale' theft by using proxy accounts and jailbreaking techniques to extract the capabilities of US artificial intelligence (AI) models. This is interpreted as a strong determination to block Chinese companies from unauthorized copying or modifying of advanced US technology.
America's AI global leadership depends on our ability to export our AI to allies. We will focus resources on effectively implementing the President's export guidance and solidifying the position of American AI.
On March 16, 2026, the Department of Commerce announced additional implementation plans for the 'American AI Exports Program.' This follows President Donald Trump's 2025 AI Action Plan and export guidance, and has begun receiving export proposals for full-stack AI technology packages from industry-led consortia.
New AI Export Regime and Enhanced Security
On April 1, 2026, the Department of Commerce officially issued a Request for Proposals (RFP) to promote the overseas adoption of American AI. The program centers on creating a catalog of AI tools granted special export status by the federal government for sale to friendly nations, aiming for global standardization of US technology.
- Receiving export proposals for full-stack AI packages from US industry-led consortia
- Building a catalog of AI tools granted special export status by the federal government
- Prioritizing technology supply to allies and strengthening security protocols
These measures are a direct response to the technology theft controversy that arose after China's DeepSeek released the R1 model in January 2025. At the time, OpenAI and Microsoft strongly protested, claiming DeepSeek stole components of ChatGPT, and sent letters to the Chinese committee.
Law enforcement agencies are also moving quickly. On April 17, 2026, the Eastern District of New York indicted former executives of AI technology company iLearningEngines, and the FBI officially documented the surge in AI use by cybercriminals for the first time in a report released on April 6. Additionally, on March 20, Chinese and American nationals were indicted on charges of conspiring to smuggle AI technology.
Currently, the most advanced US models, Gemini 3.1 Pro and GPT-5.4, lead the market, but the technology gap with China has narrowed to just 2.7%. Experts emphasize the importance of security, analyzing that while the US maintains a computing advantage, China's algorithmic progress and regulatory flexibility could further close this gap.
Some are raising concerns about the potential misuse of AI not only by state-level threats but also by non-state actors and third countries. Major institutions like the Brookings Institution point out the urgent need for practical measures against security risks that could arise when US and Chinese technologies are combined and deployed.
In the future, the US government plans to more strictly monitor compliance with export controls on semiconductors and AI chips. According to a report released by the Government Accountability Office (GAO) on March 21, 2026, the Department of Commerce has already begun phased steps to implement advanced semiconductor rules and address regulatory compliance challenges, which are expected to be a key pillar of AI model protection in the future.
This content is for information and commentary only and is not investment advice.
Join the reader conversation
Read reactions to this article and leave your own note.