White House accuses China of ‘industrial scale’ theft of AI technology

April 23, 2026

```json { "title": "White House Accuses China of 'Industrial-Scale' AI Theft", "metaDescription": "The White House says China is stealing U.S. AI technology at industrial scale. Here's what the evidence shows, from chip smuggling to model distillation.", "content": "<h2>White House Accuses China of 'Industrial-Scale' AI Technology Theft</h2>\n\nThe Trump administration has escalated its accusations against China over artificial intelligence espionage, with White House science and technology chief Michael Kratsios warning that Chinese entities are stealing American AI capabilities across multiple fronts — from adversarial model distillation to GPU smuggling rings and insider espionage at leading U.S. labs. The accusations, backed by a growing body of prosecutions, corporate disclosures, and congressional testimony, mark a significant hardening of Washington's posture on AI as a national security issue.\n\n<h2>The Distillation Problem: How Chinese Firms Extracted U.S. AI Capabilities</h2>\n\nOne of the most technically novel vectors of AI theft involves a technique known as adversarial distillation — a process in which competitors systematically query a frontier AI model with massive volumes of prompts in order to replicate its capabilities and train rival systems at a fraction of the development cost. The scale at which this has allegedly occurred is striking.\n\nIn February 2026, Anthropic published findings identifying three Chinese AI companies — DeepSeek, Moonshot AI, and MiniMax — as having conducted what it described as coordinated extraction campaigns against its Claude model. According to the company's statement, as reported by Fortune on February 24, 2026, the three firms collectively generated over 16 million exchanges with Claude through approximately 24,000 fraudulent accounts, in violation of Anthropic's terms of service and regional access restrictions.\n\nThe breakdown of that activity is revealing. MiniMax alone accounted for more than 13 million of those fraudulent exchanges. Moonshot AI generated over 3.4 million exchanges. DeepSeek, while responsible for fewer than 150,000 exchanges, specifically targeted Claude's reasoning capabilities — a technically significant focus area given the competitive importance of reasoning performance in frontier AI systems.\n\nAnthropic was unambiguous in characterizing the activity. "We have identified industrial-scale campaigns by three AI laboratories — DeepSeek, Moonshot, and MiniMax — to illicitly extract Claude's capabilities to improve their own models," the company stated, as reported by Fortune.\n\nAnthropic was not the only target. Google's Threat Intelligence Group separately disclosed that it identified and disrupted distillation and model extraction attacks aimed at its Gemini model's reasoning capabilities — attacks conducted through more than 100,000 prompts, according to reporting by The Hacker News on February 28, 2026. And according to Just Security, in early 2025 Microsoft security researchers observed individuals allegedly affiliated with DeepSeek exfiltrating large volumes of data through the OpenAI API. When DeepSeek's R1 model launched shortly thereafter, White House AI adviser David Sacks stated publicly that there was "substantial evidence" that DeepSeek had distilled from OpenAI's models.\n\nDmitri Alperovitch, chairman of the Silverado Policy Accelerator and co-founder of CrowdStrike, testified before the House Select Committee on the CCP on April 16, 2026, putting the distillation problem bluntly: "It's been clear for a while now that part of the reason for the rapid progress of Chinese AI models has been theft via distillation of U.S. frontier models."\n\nIn response to the escalating threat, OpenAI, Anthropic, and Alphabet announced on April 6, 2026 that they would begin sharing threat intelligence related to Chinese-linked industrial espionage through the Frontier Model Forum — a non-profit they co-founded with Microsoft in 2023. According to the Foundation for Defense of Democracies, this marked the first coordinated security operation among the companies specifically targeting AI theft.\n\n<h2>Chip Smuggling and Insider Espionage: The Hardware and Human Dimensions</h2>\n\nThe distillation campaigns represent only one layer of what U.S. officials describe as a multi-vector campaign to close the AI capability gap. Two additional fronts — hardware smuggling and insider espionage — have produced a wave of major prosecutions over the past several months.\n\nOn the hardware side, the scale of alleged violations has been extraordinary. In December 2025, the Department of Justice publicly announced "Operation Gatekeeper," which dismantled a China-linked smuggling network operating across the United States, Singapore, Malaysia, and Hong Kong. According to Alperovitch's written testimony to the House Select Committee on the CCP, the operation involved at least $160 million worth of Nvidia H100 and H200 GPUs.\n\nThat case was followed in March 2026 by an indictment unsealed in Manhattan federal court charging Yih-Shyan "Wally" Liaw, co-founder of Super Micro Computer, along with two others, with allegedly working to divert billions of dollars in Supermicro AI servers to China in violation of export controls. According to the House Select Committee on the CCP, the case has been described by Committee Chairman John Moolenaar as potentially the largest export control violation in U.S. history, with the alleged diversion valued at approximately $2.5 billion.\n\nChairman Moolenaar was direct in his assessment of the broader smuggling problem: "China's smuggling of advanced AI chips is a pervasive threat facing law enforcement," he stated in his opening remarks at the April 16 hearing titled "China's Campaign to Steal America's AI Edge."\n\nThe hardware smuggling cases exist against a backdrop of surging Chinese investment in chip manufacturing infrastructure. According to written testimony submitted to the House Select Committee on the CCP, China imported a record $51.1 billion of semiconductor manufacturing equipment during 2025 — up from just $10.7 billion in 2016. That nearly fivefold increase in a decade underscores Beijing's determination to reduce its dependence on American semiconductor supply chains regardless of what controls Washington imposes.\n\nOn the insider espionage front, the Department of Justice secured what the Foundation for Defense of Democracies described as a landmark milestone in January 2026: the first-ever conviction on economic espionage charges related to AI. Former Google software engineer Linwei Ding, a Chinese national, was convicted on 14 counts of economic espionage and trade-related theft. According to congressional testimony submitted to the House Select Committee on the CCP, the jury found that in 2022 and 2023, Ding had stolen more than 2,000 pages of Google's AI-related trade secrets — thefts that were sparked, the testimony notes, by an encounter with Chinese intelligence officers.\n\nThe Ding case is unlikely to be isolated. According to testimony submitted to the April 16, 2026 House Select Committee hearing, approximately 38% of top AI researchers at American AI labs and research institutions received their undergraduate education in China, and the vast majority of those researchers are likely Chinese nationals. That statistic has drawn significant attention in Washington as policymakers grapple with how to protect sensitive AI research without undermining the research talent pipeline that has helped sustain American AI leadership.\n\n<h2>The Strategic Stakes: America's AI Lead Is Narrowing</h2>\n\nUnderlying all of these specific incidents is a broader strategic concern: that the cumulative effect of distillation, hardware acquisition, and insider access is allowing China to compress what was once a meaningful AI capability gap far more rapidly than it could through organic research and development alone.\n\nMichael Kratsios, who serves as Director of the White House Office of Science and Technology Policy and was confirmed in that role on March 25, 2025, has been the administration's most prominent voice on this issue. Testifying before Congress in January 2026, he acknowledged the trajectory plainly: "In 2020, the American innovation enterprise held a comfortable lead in AI. By 2024, that gap had begun to close significantly."\n\nAccording to congressional testimony submitted to the House Select Committee hearing on April 16, 2026, China's frontier AI capabilities are currently estimated to be approximately seven months behind America's. Whether that gap continues to close — and at what rate — is a central preoccupation for U.S. national security planners.\n\nKratsios has articulated why the global AI infrastructure question matters beyond pure military or economic competition. Speaking at the Center for Strategic and International Studies, he warned: "If most countries around the world are running on an AI stack that isn't American, and potentially one's an adversary, that's a really, really big problem." The concern is not only about direct theft of U.S. capabilities, but about which country's technological ecosystem becomes the default foundation for AI development globally.\n\nThe White House AI Action Plan — drafted by the Kratsios-led OSTP — outlines a three-pillar strategy built around accelerating innovation, building critical infrastructure, and strengthening U.S. influence abroad, according to Kharon. Export controls on advanced semiconductors are a core component of that strategy. As Kratsios stated at CSIS: "The highest end of semiconductors need to continue to be export controlled, not allowed into China," and any sale of Nvidia chips to China "will require an export license."\n\nBut Kratsios has also been candid about the limitations of export controls as a policy instrument on their own. "You can have the best export controls in the books, but if you're not able to effectively enforce them because of resource constraints, that's a challenge," he said at CSIS — a statement that takes on added weight given the scale of the smuggling operations that law enforcement has since uncovered.\n\nThe FBI has similarly flagged the AI espionage threat. According to VOA News, a senior FBI official stated: "Nation-state adversaries, particularly China, pose a significant threat to American companies and national security by stealing our AI technology and data to advance their own AI programs."\n\n<h2>What Comes Next</h2>\n\nThe April 16, 2026 House Select Committee hearing on "China's Campaign to Steal America's AI Edge" represented the most comprehensive congressional examination to date of how China is acquiring advanced U.S. AI technology through both legal and illicit channels. The hearing is expected to inform legislative action on export control enforcement, AI lab security requirements, and researcher vetting protocols — though specific legislation had not been enacted as of the date of publication.\n\nThe coordination agreement among OpenAI, Anthropic, and Alphabet through the Frontier Model Forum represents a new operational posture for the U.S. AI industry — one in which leading labs treat Chinese distillation campaigns as a shared threat requiring collective defense, rather than a company-specific compliance problem. Whether that coordination produces measurable reductions in successful extraction attacks remains to be seen.\n\nOn the hardware enforcement side, the Super Micro indictment and the Ding conviction signal that the DOJ is prepared to pursue high-profile prosecutions in the AI espionage space. But as Kratsios himself acknowledged, enforcement capacity relative to the scale of smuggling activity remains a structural constraint.\n\nChina's record $51.1 billion in semiconductor manufacturing equipment imports in 2025 also suggests that Beijing is pursuing a parallel strategy: accelerating domestic chip production capability so that U.S. export controls become progressively less decisive over time. The race between American AI capability leadership and Chinese efforts to close that gap — through whatever means are available — is shaping up to be one of the defining technology competitions of the decade.\n\nFor more tech news, visit our <a href=\"/news\">news section</a>.\n\n<h2>Why This Matters for Productivity and Performance</h2>\n\nThe battle over AI technology isn't just a geopolitical story — it shapes which AI tools you use, how trustworthy they are, and whether the platforms powering your work and health decisions are built on a foundation of genuine innovation or compromised development. Staying informed about the forces shaping AI helps you make smarter decisions about the technology you rely on every day. <a href=\"/#waitlist\">Join the Moccet waitlist to stay ahead of the curve.</a>", "excerpt": "The Trump administration has accused Chinese entities of stealing U.S. AI technology at industrial scale, pointing to adversarial model distillation campaigns, GPU smuggling rings, and insider espionage. Anthropic reported that three Chinese AI firms generated over 16 million fraudulent exchanges with its Claude model, while the DOJ secured its first-ever AI economic espionage conviction and unsealed a $2.5 billion export control case. A House Select Committee hearing convened on April 16, 2026 to examine the full scope of China's campaign to acquire American AI capabilities.", "keywords": ["AI theft China", "industrial-scale AI espionage", "Michael Kratsios OSTP", "DeepSeek distillation", "AI export controls"], "slug": "white-house-china-industrial-scale-ai-theft" } ```

← Back to Tech News

More Tech News

U.S. Accuses China of Industrial-Scale AI Theft

A group of users leaked Anthropic’s AI model Mythos by reportedly guessing where it was located

Era Computer Raises $11M for AI Wearables Platform