White House accuses China of copying American AI models in ‘industrial-scale’ campaign

The White House has accused foreign entities, primarily based in China, of conducting "industrial-scale" campaigns to steal proprietary AI models from U.S. companies, according to a memo from Michael Kratsios, director of the Office of Science and Technology Policy.

The campaigns involve tens of thousands of surrogate accounts using sophisticated tools to extract proprietary information from American AI systems, Kratsios wrote. "These coordinated campaigns systematically extract capabilities from American AI models, exploiting American expertise and innovation."

The memo represents the most specific public accusation yet from the administration regarding AI-related intellectual property theft. While cyber-espionage from China has been a persistent concern across defense and technology sectors, the framing of AI model extraction as an "industrial-scale" operation elevates the issue from corporate security to national strategic priority.

The methods described go beyond simple hacking. The campaigns reportedly use mass automated queries to probe model behaviors, reverse-engineer training data, and expose safety guardrails — effectively creating copies of proprietary models without direct code theft. This is a newer form of IP extraction that existing trade secret laws weren't designed to address.

The accusation comes amid an intensifying U.S.-China race for AI supremacy. Both nations have invested hundreds of billions in AI infrastructure, with the U.S. maintaining a lead in frontier models but China closing the gap rapidly — partly, the memo implies, through these extraction campaigns.

The administration is expected to announce new restrictions on AI model access and export controls in the coming weeks, building on previous executive orders limiting the sale of advanced AI chips to Chinese entities.

**What This Means For You:** If you work in AI or tech, expect tighter controls on model access, API usage, and cross-border data flows. Companies deploying AI will face new compliance requirements around model security and access logging. For investors, this accelerates the thesis that AI infrastructure with strong security moats — closed-source models with robust access controls — will command premium valuations. Open-weight models may face new regulatory headwinds.