White House Memo Accuses China of Industrial-Scale AI Theft 2026

Published: 24 April 2026. The English Chronicle Desk. The English Chronicle Online

In a move that has sent shockwaves through the global tech sector, the White House has issued a blistering memo accusing Chinese firms of conducting “industrial-scale” theft of American artificial intelligence. The directive, authored by Michael Kratsios, Director of the Office of Science and Technology Policy, alleges that foreign entities—primarily based in China—are engaged in coordinated campaigns to “distill” and systematically extract the proprietary capabilities of U.S. frontier AI models.

The timing of the memo is as significant as its content. Released on Thursday, April 23, the document arrives just weeks before a high-stakes summit between President Donald Trump and President Xi Jinping in Beijing. The accusations threaten to shatter the fragile “tech detente” brokered last October and have already triggered a bipartisan push in Congress for aggressive new sanctions against the firms involved.

According to the White House, the theft is being carried out through a sophisticated process known as “model distillation.” While legitimate distillation is a common practice used to create smaller, more efficient versions of AI models, the memo claims Chinese actors are using it as a tool for “surreptitious exploitation.”

The Proxy Army: The memo reveals that the U.S. government has detected the use of tens of thousands of proxy accounts designed to evade the rate limits and security filters of American AI labs.

Jailbreaking Techniques: Coordinated campaigns are reportedly using “jailbreaking” prompts to bypass safety protocols, forcing models to reveal proprietary information about their underlying architecture and training data.

The Competitive Shortcut: By “distilling” American models, Chinese firms can release products that mimic the performance of top-tier US AI—such as OpenAI’s GPT or Anthropic’s Claude—at a fraction of the original R&D cost.

While the memo did not name specific companies, it follows months of mounting tension between U.S. labs and their Chinese counterparts. In February 2026, Anthropic publicly accused three leading Chinese firms—DeepSeek, Moonshot AI, and MiniMax—of running distillation attacks on its Claude chatbot. Similarly, OpenAI has previously alleged that DeepSeek used GPT outputs to train its own models in violation of its terms of service.

A spokesperson for the Chinese Embassy in Washington dismissed the White House claims as “pure slander” and “unjustified suppression,” asserting that Beijing remains committed to the protection of intellectual property.

The Trump administration has vowed to move from accusation to action. The memo outlines a new strategy to share real-time tactical intelligence with American AI companies to help them harden their defenses. This aligns with the recently launched Project Glasswing, a multi-company initiative involving titans like NVIDIA, Microsoft, Google, and Amazon, specifically designed to secure the AI supply chain.

However, the political fallout may be the most immediate consequence. On Wednesday, the House Foreign Affairs Committee passed a series of bills that would require the administration to consider adding any group engaged in “unauthorized distillation” to the Entity List—effectively a death sentence for their access to American chips and software.

As the “AI arms race” enters this volatile new phase, the White House memo makes one thing clear: the days of treating AI distillation as a purely academic exercise are over. In the eyes of Washington, the “Silicon Curtain” is no longer just about stopping the export of chips—it is about stopping the theft of the “intelligence” those chips were built to create.

Check out Our Latest News