How China Is Stealing America’s AI

Read the original at The Free Press ↗
Cite

Pottinger, M. (2026, September 9). How China Is Stealing America’s AI. The Free Press. https://www.thefp.com/p/china-america-ai-race

Dated when The Aporia collected the article, usually within hours of publication. Check the original for its exact date and byline.

The Free Press · collected 2026-09-09 · by Matt Pottinger

Quick Summary

A Chinese tech expert describes China's rapid advancements in artificial intelligence (AI) as being largely due to "stealing" from American companies, rather than solely through innovation. The article highlights that large US corporations are now using Chinese AI models such as DeepSeek and Qwen, which were previously unknown in the US business world. According to the expert, China's success is partly attributed to a technique called distillation, where a state-of-the-art American model is used to train a new one by asking millions of questions and recording its answers. This practice, done with permission from the original model owners, is considered acceptable research in some cases, such as Apple's use of Google's Gemini model to improve Siri.
Written locally by qwen2.5:14b on 2026-09-09, using this article's own text rather than the other coverage of the same event (that is the story summary below).

AI analysis runs on qwen2.5:14b, locally

Story summary

The US has accused China of stealing its artificial intelligence (AI) secrets, including a technique called "distillation", which allows smaller AI systems to be trained using outputs from larger models. The US Treasury secretary warned that if China wins the AI race, it will have a significant impact on global competition for technological supremacy. Meanwhile, the transportation secretary expressed concern about Ford's ties to Chinese firms that he believes are detrimental to the US automotive industry. In response, China has rejected the allegations, calling them "unfounded accusations" and urging the US to strengthen cooperation instead of making unfounded smears.

The US authorities have identified six Chinese entities involved in this alleged theft, including DeepSeek, Moonshot AI, and Alibaba, which they claim are tapping into American AI products to accelerate their own development. The technique of distillation allows these smaller systems to be trained using outputs from larger models, drastically lowering the cost of building new tools.

The issue is expected to be a major topic at a planned meeting between US President Donald Trump and Chinese President Xi Jinping later this month, where AI governance is expected to be discussed. China's foreign ministry spokesperson Mao Ning has said that China's progress in AI is driven by "achieving high-level science and technological self-reliance", and urged the US to earnestly implement agreements reached by leaders of both countries.

Written for “US vs China AI Theft Accusations” on 2026-10-24, grounded in this article and the 3 other(s) covering the same event.

Story

📰 US vs China AI Theft Accusations
Politics · 4 article(s) covering the same event. Compare every outlet's account →

Compared with similar articles

This article reads leans right and hedges 0% of its claims. Each row says how that neighbour differs.
Feds accuse China of industrial-scale theft of America’s AI secrets same event · 100%
The Independent
⚖️ leaning not scored 🔴 27% hedged 3 of 11 📰 publisher trust 59
“Both articles are dated September 9th, 2026 and mention a joint statement issued by U.S. intelligence and law enforcement officials accusing Chinese AI firms of industrial-scale theft of America's AI secrets.”
AI that works for Americans different event · 90%
Washington Examiner
⚖️ Leans left further left than this 🔴 5% hedged 3 of 55 📰 publisher trust 66
“Article A discusses the emergence of AI in general and Americans' trust issues with it, while Article B focuses on China's alleged theft of American companies' AI advancements”
Fox News AI Newsletter: The network driving fight against data centers, Flock cameras different event · 90%
Fox News
⚖️ leaning not scored 🔴 14% hedged 3 of 21 📰 publisher trust 67
“Article A refers to a protest network called 'No Kings' while Article B does not mention protests or specific events”
U.S., China gear up for mid-September AI safety talks different event · 90%
The Hindu
⚖️ leaning not scored 🔴 10% hedged 4 of 39 📰 publisher trust 60
“Article A reports on planned talks between U.S. and China about AI safety, while Article B discusses Chinese companies allegedly stealing American AI advancements, without mentioning any specific event or meeting”
OpenAI's new Astra model has private thoughts different event · 90%
Semafor
⚖️ Leans left further left than this 🔴 0% hedged 0 of 5 📰 publisher trust 60
“Article A discusses a new AI model called Astra, while Article B mentions Chinese AI models like DeepSeek, Moonshot's Kimi K3, and Alibaba Cloud's Qwen without referencing Astra”
Punching above their weight: how China’s AI giants stretch each dollar in compute race different event · 90%
South China Morning Post
⚖️ Leans left further left than this 🔴 25% hedged 1 of 4 📰 publisher trust 64
“Article B mentions a general trend of China's open-weight AI models, while Article A discusses a specific economic analysis by Moody's on the cost savings of Chinese tech firms in the compute race”
‘Unfounded accusations’: China rejects US claims of 'malicious' AI distillation by its firms different event · 90%
Times of India
⚖️ leaning not scored 🔴 15% hedged 2 of 13 📰 publisher trust 59
“Article A describes a specific response from China to US allegations, while Article B discusses a broader topic of Chinese AI companies 'stealing' American capabilities, without mentioning any specific incident or date”
AI researchers say industry is 'gambling with our lives' different event · 90%
Semafor
⚖️ leaning not scored 🔴 29% hedged 2 of 7 📰 publisher trust 60
“The articles cover different topics, with Article A discussing concerns about AI development and potential risks, while Article B discusses China's alleged theft of American AI”
Why the AI hyperscalers are winning the data center debate different event · 90%
Washington Examiner
⚖️ Leans strongly right further right than this 🔴 4% hedged 1 of 26 📰 publisher trust 66
“Article A discusses a trend of Chinese AI models competing with American ones through theft, while Article B reports on hyperscalers like Oracle and Meta pushing back against resistance to data centers in northern Virginia”
Anthropic Staffers Again Sound the Alarm on AI Catastrophe different event · 90%
Mother Jones
⚖️ leaning not scored 🔴 18% hedged 8 of 44 📰 publisher trust 95
“Article A discusses the threat of Chinese open-weight models, while Article B focuses on concerns raised by Anthropic staffers about AI catastrophe and superintelligence”
How this article was measured signals, leaning charts, publisher record, byline, topics and claims

Signals How these are calculated →

Claims extracted
10
sentences that state something as happening or true
Uncertain
0%
0 of 10 hedged ("reportedly", "may", "according to")
Leaning
Leans right
of the writing, not the subject · beta estimate
Publisher signals
96/100
The Free Press's corrections and hedging, across everything collected from it. Not a measure of accuracy.
Outlets on this story
4
Politics
Same framing elsewhere
1
articles that frame the event the way this one does
Analyzed 2026-09-09 · how these are computed

How this is being covered How these are calculated →

Article leaning vs. publisher reliability
Source leaning vs. consistency

Publisher

The Free Press · 357 article(s) · 0 correction(s) detected
No corrections detected for this publisher. That may mean careful reporting, or simply that nothing has been checked.

Who wrote this

Matt Pottinger
1 article(s) here · 1 carrying a prediction
🔮 American companies will spend roughly $700 billion this year on the physical build-out of artificial intelligence: data centers, chips, and computing power.
2026-09-09 · assertive framing · How China Is Stealing America’s AI
The only article under this byline in the corpus.

Topics

American Chinese DeepSeek Moonshot U.S.

Subjects

American NORP · 4× Chinese NORP · 2× Airbnb ORG · 1× Alibaba Cloud’s ORG · 1× Coinbase ORG · 1× DeepSeek ORG · 1× DoorDash ORG · 1× Kimi K3, PERSON · 1× Moonshot ORG · 1× U.S. GPE · 1×

Narrative

It also owes much to a technique called distillation, in which a state-of-the-art American model is asked millions of questions, its answers are recorded, and a new model is trained on them—inheriting much of the original’s capability without needing anything like the research and investment that produced it.
framing: assertive · carried by 1 article(s) · first seen 2026-09-09
🔮 American companies will spend roughly $700 billion this year on the physical build-out of artificial intelligence: data centers, chips, and computing power.
2026-09-09 · The Free Press
How China Is Stealing America’s AI · assertive framing

Claims (10 extracted, 0 hedged)

American companies will spend roughly $700 billion this year on the physical build-out of artificial intelligence: data centers, chips, and computing power. asserted
companies → spend → intelligence
Every dollar is a bet on American capitalism—that the advances those facilities help produce will earn enough to fund the next breakthrough. asserted
facilities → help → breakthrough
The advent of Chinese open-weight models is undermining that bet, not just by ordinary competition, but by theft. asserted
advent → undermine → theft
(Open-weight means its code is freely downloadable and customizable, unlike its proprietary U.S. rivals.) asserted
code → mean → rivals
Chinese AI models like DeepSeek, Moonshot’s Kimi K3, and Alibaba Cloud’s Qwen barely registered in American C-suites before last year. asserted
models → register → year
They now run inside Airbnb, Coinbase, DoorDash, and many others. asserted
They → run → Airbnb
It also owes much to a technique called distillation, in which a state-of-the-art American model is asked millions of questions, its answers are recorded, and a new model is trained on them—inheriting much of the original’s capability without needing anything like the research and investment that produced it. asserted
that → owe → it
Done with permission, it’s an acceptable research practice. asserted
it → do → permission
Apple pays Google to distill the latter’s Gemini model and make a smarter Siri. asserted
Apple → pay → Siri
University labs use the technique to study how models work. asserted
models → use → technique
💬Give feedback
🕘History 🎫Support
Terms · Privacy · beta v2.0.0
Articles to 24 Oct 2026