US Agencies Accuse China of 'Industrial-Scale' Theft of American AI Technology

.

A Joint Warning From Washington's Top Security Agencies

The US Cybersecurity and Infrastructure Security Agency (CISA), the National Security Agency (NSA), and the FBI jointly published a cybersecurity advisory on September 8, 2026, accusing China-based AI firms of running a coordinated campaign to copy American AI technology. The agencies described the activity as "aggressive, malicious and targeted distillation" carried out "at an industrial scale."

Distillation itself is not illegal. It is a standard machine-learning method in which a smaller, cheaper AI model is trained using the outputs of a larger, more advanced one. What the US agencies allege is different: that Chinese firms used this technique to extract restricted, proprietary capabilities from American systems, in ways that violate the terms of use of those platforms.

.

Six Companies Named

The advisory names six Chinese firms: DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI. According to the agencies, these companies targeted leading US models, including systems built by Anthropic, OpenAI, Google, and Elon Musk's xAI, sending enormous volumes of queries to harvest their outputs.

DeepSeek is accused of running an organized campaign since at least late 2024 to train its R1 and V3 models, focusing on reasoning ability and specialized functions. Moonshot AI reportedly used millions of queries against roughly 18 different US models to build its Kimi-K2 and Kimi-K3 systems. Alibaba is said to have used similar methods to improve its Qwen model family.

To avoid detection, the agencies say the Chinese firms spread their queries across thousands of accounts, cloud providers, and third-party services — a method described in the advisory as "distributed distillation," designed so that no single account would generate enough traffic to raise alarms.

.

Not Just Business — A National Security Concern

US officials stressed that this is not merely a commercial dispute over technology. The advisory states the practice strengthens China's "military and cyberattack capabilities that could be used against the U.S. and our allies." A separate Reuters report from July 31, 2026 found that Chinese military researchers have already used outputs from leading US AI models to help train domestic defense systems.

The advisory was coordinated with intelligence partners in the United Kingdom, Australia, Canada, and New Zealand — the so-called "Five Eyes" alliance — underlining how seriously Washington and its allies are treating the issue.

.

Timing Matters: A Sensitive Moment in US-China Relations

The accusation lands at a delicate diplomatic moment. Chinese President Xi Jinping is due to visit the United States in late September, and officials from both governments were already preparing for a mid-September dialogue specifically on AI safety risks. The Trump administration made a similar accusation back in April 2026, shortly before Trump's own visit to Beijing — suggesting Washington is willing to apply public pressure ahead of high-level diplomacy rather than raise the issue only behind closed doors.

The Chinese embassy in Washington did not immediately respond to requests for comment on the new allegations.

.

Context: A Pattern, Not an Isolated Incident

This is not the first time Chinese AI firms have faced such accusations. DeepSeek's rapid rise in early 2025 already triggered speculation in the US tech industry that its models had been trained using outputs from OpenAI's systems. Tuesday's advisory is the most detailed and formal accusation to date, naming specific companies, specific target models, and specific techniques used to evade detection.

Some independent AI researchers note that distillation is common industry practice worldwide, and that the White House itself acknowledged in an April 2026 memo that distilled models "do not replicate the full performance of the original," even if they can look competitive on certain benchmarks. This raises a fair question about how much of a genuine technological edge China has actually gained versus simply narrowing the gap on paper.

.

What Comes Next

CISA's advisory recommends that US AI companies strengthen detection systems, monitor unusual usage patterns, and consider subtly altering outputs to reduce the value of unauthorized distillation attempts. Whether these recommendations will be paired with concrete trade or diplomatic consequences remains to be seen — especially with the Trump-Xi meeting just weeks away.

For now, the episode adds another point of friction to an already tense technological rivalry between the world's two largest economies, one in which Washington continues to position itself as defending both its companies' intellectual property and its broader strategic lead in artificial intelligence.

.

Sources

1. Reuters – "US accuses Chinese AI firms of 'malicious' copying of AI technology" – https://www.reuters.com/technology/us-accuses-chinese-ai-firms-industrial-scale-theft-ai-technology-2026-09-08/
2. CISA (official advisory, AA26-251A) – "China-Based Artificial Intelligence Companies Conducting Industrial-Scale Distillation Campaigns Against U.S. AI Companies" – https://www.cisa.gov/news-events/cybersecurity-advisories/aa26-251a
3. CISA – Press release – "CISA, NSA and FBI Warn of China-Based AI Companies Targeting US AI Models with Industrial-Scale Knowledge Distillation Campaigns" – https://www.cisa.gov/news-events/news/cisa-nsa-and-fbi-warn-china-based-ai-companies-targeting-us-ai-models-industrial-scale-knowledge
4. NBC News – "US accuses China AI developers DeepSeek and Alibaba of copying American AI" – https://www.nbcnews.com/tech/tech-news/us-accuses-china-ai-developers-deepseek-alibaba-copying-american-ai-rcna596696
5. CNN Politics – "US claims Chinese AI firms are carrying out 'industrial-scale' theft of trade secrets" – https://edition.cnn.com/2026/09/08/politics/us-accuses-china-of-stealing-ai-technology
6. Reuters – "Chinese military researchers tap US AI models to train defence systems" (July 31, 2026) – https://www.reuters.com/world/asia-pacific/chinese-military-researchers-tap-us-ai-models-train-defence-systems-2026-07-31/

.