White House Accuses China of AI Crimes

White House Office of Science and Technology Policy Director Michael Kratsios released a memo on April 23 accusing Chinese agencies of conducting deliberate, industrial campaigns to disrupt and hack US border AI systems, using tens of thousands of proxy accounts and jailbreak methods, days before the planned Trump-Xi summit.
Summary
- The White House OSTP memo accuses Chinese enterprises primarily based in China of conducting coordinated campaigns to disperse AI systems across the US border from OpenAI, Anthropic, and other American labs.
- The memo says the attackers used tens of thousands of proxy accounts to evade detection and jailbreak techniques to expose proprietary model information to closed American AI systems.
- The administration will share intelligence with US AI companies about active targeting campaigns and explore ways to hold them accountable against responsible foreign actors.
White House OSTP director Michael Kratsios released a memo to US government agencies on April 23 saying foreign agencies “primarily based in China” are “engaged in deliberate, industrialized campaigns to dismantle US border AI systems.” The lawsuit represents the most formal and definitive statement the Trump administration has made regarding the theft of Chinese AI intellectual property from American laboratories and comes days before the planned Trump-Xi summit.
China’s White House Stealing AI Names Distillation as Attack Vector
Distillation is a technical process in which small AI models are trained on large, proprietary results, allowing the lab to measure the performance of the frontier without paying the full training costs associated with building it from scratch. According to the Kratsios memo, Chinese actors used tens of thousands of proxy accounts to make massive inquiries into American AI systems, evading ratings restrictions and corporate identification systems designed to flag suspicious usage patterns. Jailbreaking techniques are then used to strip the security filters and reveal the proprietary behavior of the models, which are captured as training data for the digested copies. Both OpenAI and Anthropic have publicly accused DeepSeek of leaking their models. The Financial Times reported that the memo was directed at US government agencies and indicates that the administration will begin sharing intelligence on effective distillation campaigns directly with US AI companies. As reported by crypto.news, Nvidia CEO Jensen Huang had warned a few days ago that China has ghost data centers with massive computing infrastructure that can match the AI capabilities of the US border.
DeepSeek’s V4 Launch Sharpens Claim Time
DeepSeek released preview versions of its V4 model on April 24, built to run on Huawei’s Ascend chips, one day after the Kratsios memo appeared. Chinese rivals Zipu AI and MiniMax fell 9% and 7% respectively on the day of the V4 launch, reading the release as a competitive threat. DeepSeek has denied allegations of illegal use of artificial data for training, insisting that its data is collected through natural web searches. China’s ambassador to Washington called the White House “unfounded” and said Beijing “is very important to the protection of intellectual property rights.” The State Department urged Washington to “stop bias.” As crypto.news is written, US AI companies including Nvidia, Microsoft, and Meta have been hovering around including country and political exposure in China throughout the year 2026, with the IRGC’s designation of American technology companies as targets in early April adding a second layer of pressure on top of the alleged theft.
What the Administrators Say Will Do
The memo says the administration will share intelligence about distillation campaigns with American AI companies and “explore more ways to hold foreign actors accountable.” Separately, Commerce Secretary Howard Lutnick confirmed on April 23 that no shipments of Nvidia’s advanced AI chip had arrived in China despite January’s conditional approval, adding the size of the chip supply to a dispute already being fought over model and data levels. The question of whether the Trump-Xi summit scheduled for the coming days will directly address the intellectual property of AI is being closely watched by industry executives, given that both the alleged distillation and the dispute over the supply of chips are now contentious issues between the two governments. As crypto.news tracked, Nvidia’s income has already been affected by the tightening of chip export limits that increased from 2025 to 2026, making the ambassador’s decision on AI trade a major impact on the US semiconductor industry.
The White House memo did not name specific Chinese companies or individuals, nor did it announce any sanctions, fines, or executive actions against organizations suspected of food laundering campaigns.



