U.S. accuses China of "industrial-scale" campaigns to steal AI secrets
The Trump administration accused China-backed actors of conducting industrial-scale campaigns to copy American AI models by querying proprietary APIs. A memo from the White House Office of Science and Technology Policy detailed these 'distillation attacks,' warning that such tactics could strip away guardrails meant to maintain ideological neutrality. The administration plans to share intelligence with U.S. AI companies to help develop defenses.
open_in_new
Read the original article: https://axios.com/2026/04/23/us-china-ai-theft-distillation
analyticsAnalysis
30%
Propaganda Score
confidence: 90%
Minor concerns. Some persuasive language detected, but largely factual.
psychologyDetected Techniques
warning
Loaded Language
80% confidence
Using words with strong emotional connotations to influence an audience.
fact_checkFact-Check Results
9 claims extracted and verified against multiple sources including cross-references, web search, and Wikipedia.
check_circle
Corroborated
5
info
Single Source
4
“The Trump administration on Thursday accused China-backed actors of running "deliberate, industrial-scale campaigns" to distill and copy American frontier AI models.”
CORROBORATED
Multiple web search results report that the Trump administration/White House accused China of running 'industrial-scale campaigns' to steal U.S. AI technology. The evidence points to the White House making these accusations.
menu_book
wikipedia
NEUTRAL
— Throughout both of his presidencies, U.S. president Donald Trump has expressed a desire to expand the United States' territory and influence through both land purchases and military means.
Trump first…
https://en.wikipedia.org/wiki/American_expansionism_under_Do…
https://en.wikipedia.org/wiki/American_expansionism_under_Do…
menu_book
wikipedia
NEUTRAL
— Anthropic PBC is an American artificial intelligence (AI) company headquartered in San Francisco. It has developed a range of large language models (LLMs) named Claude and focuses on AI safety.
Anthro…
https://en.wikipedia.org/wiki/Anthropic
https://en.wikipedia.org/wiki/Anthropic
menu_book
wikipedia
NEUTRAL
— "Trump Always Chickens Out" (TACO) is a pejorative description of the perceived tendency of United States president Donald Trump to make threats, only to later delay or renege on them as a way to incr…
https://en.wikipedia.org/wiki/Trump_Always_Chickens_Out
https://en.wikipedia.org/wiki/Trump_Always_Chickens_Out
+ 3 more evidence sources
“Michael Kratsios, director of the White House Office of Science and Technology Policy, sent a memo Thursday to federal agency heads accusing mostly China-based actors of using proxy accounts to evade detection and jailbreak models to "expose proprietary information" and "extract capabilities from American AI models."”
CORROBORATED
Multiple web search results confirm that Michael Kratsios sent a memo accusing China-based actors of using proxy accounts and jailbreaking models to expose proprietary information and extract capabilities.
travel_explore
web search
NEUTRAL
— Driving the news: Michael Kratsios, director of the White House Office of Science and Technology Policy, sent a memo Thursday to federal agency heads accusing mostly China-based actors of using proxy …
https://www.axios.com/2026/04/23/us-china-ai-theft-distillat…
https://www.axios.com/2026/04/23/us-china-ai-theft-distillat…
travel_explore
web search
NEUTRAL
— China slams accusations as “pure slander”. Kratsios’ memo threatening a crackdown comes ahead of Donald Trump’s highly anticipated meeting with China’s president Xi Jinping next month. Trump has claim…
https://arstechnica.com/tech-policy/2026/04/us-accuses-china…
https://arstechnica.com/tech-policy/2026/04/us-accuses-china…
travel_explore
web search
NEUTRAL
— He noted that actors use proxy accounts to avoid detection while probing AI systems. He also said they use jailbreaking techniques to expose restricted or proprietary outputs. Officials believe these …
https://interestingengineering.com/ai-robotics/ai-war-white-…
https://interestingengineering.com/ai-robotics/ai-war-white-…
“Distillation attacks involve querying proprietary models, like Claude or Gemini, millions of times via APIs to build datasets that replicate how the systems behave.”
CORROBORATED
Multiple web search results define distillation attacks as involving querying proprietary models via APIs to reconstruct or replicate the target's behavior, which aligns with the claim's description.
travel_explore
web search
NEUTRAL
— Distillation attacks represent an existential threat to the current AI business model. Companies invest hundreds of millions in training frontier models, expecting to monetize through API access.
https://mdpgroup.com/en/blog/distillation-attacks-industrial…
https://mdpgroup.com/en/blog/distillation-attacks-industrial…
travel_explore
web search
NEUTRAL
— A distillation attack occurs when an adversary interacts with a machine learning model (typically via API) and reconstructs a substitute model that closely approximates the target’s behavior.
https://www.linkedin.com/pulse/distillation-attacks-ratnadee…
https://www.linkedin.com/pulse/distillation-attacks-ratnadee…
travel_explore
web search
NEUTRAL
— Models built through illicit distillation are unlikely to retain those safeguards, meaning that dangerous capabilities can proliferate with many protections stripped out entirely. Foreign labs that di…
https://www.anthropic.com/news/detecting-and-preventing-dist…
https://www.anthropic.com/news/detecting-and-preventing-dist…
“Kratsios said these campaigns enable foreign actors to release models that appear to match U.S. AI capabilities at a fraction of the cost.”
SINGLE SOURCE
The claim that Kratsios stated foreign actors can release models matching U.S. capabilities at a fraction of the cost is supported by general context in the web search results regarding the threat, but the exact phrasing and attribution to Kratsios are not independently corroborated by multiple sources.
travel_explore
web search
NEUTRAL
— Michael John Kotsakas Kratsios (born November 7, 1986) [1] is an American business executive and government official, who has served as the 13th director of the White House Office of Science and Techn…
https://en.wikipedia.org/wiki/Michael_Kratsios
https://en.wikipedia.org/wiki/Michael_Kratsios
travel_explore
web search
NEUTRAL
— The White House on Thursday accused foreign entities, primarily China, of running "industrial-scale" campaigns to distill frontier artificial intelligence models from U.S. companies.
https://www.msn.com/en-us/news/politics/white-house-official…
https://www.msn.com/en-us/news/politics/white-house-official…
travel_explore
web search
NEUTRAL
— Highlights as Michael Kratsios, an architect of Trump's AI action plan, testifies before Congress about Grok, federal preemption and more.
https://www.techpolicy.press/transcript-ostp-director-kratsi…
https://www.techpolicy.press/transcript-ostp-director-kratsi…
“He added that such tactics can also strip away guardrails meant to keep outputs "ideologically neutral and truth-seeking."”
SINGLE SOURCE
The specific detail that Kratsios stated these tactics strip away guardrails meant to keep outputs 'ideologically neutral and truth-seeking' is mentioned in the context of the threat, but this exact phrasing is not independently corroborated by multiple sources.
travel_explore
web search
NEUTRAL
— White House Office of Science and Technology Policy Director Michael Kratsios testifies during a Senate Commerce Subcommittee on Science, Manufacturing, and Competitiveness hearing titled "AI've Got a…
https://www.techpolicy.press/white-house-office-of-science-a…
https://www.techpolicy.press/white-house-office-of-science-a…
travel_explore
web search
NEUTRAL
— Michael Kratsios 02:14 Well, first and foremost, the President wants us to win on these key, critical, and emerging technologies. We as the White House have spoken about four in particular: artificial…
https://milkeninstitute.org/sites/default/files/2025-05/conv…
https://milkeninstitute.org/sites/default/files/2025-05/conv…
travel_explore
web search
NEUTRAL
— Kratsios emphasized the need for American leadership in technology and criticized regulatory constraints on innovation. The Context of the Statement During a policy address at the Endless Frontiers Re…
https://www.intelligenthq.com/white-house-sparks-controversy…
https://www.intelligenthq.com/white-house-sparks-controversy…
“OpenAI and Anthropic both said earlier this year that China-based firms — including DeepSeek, Moonshot AI and MiniMax — were behind wide-scale distillation attacks on their models.”
CORROBORATED
Multiple web search results confirm that both OpenAI and Anthropic stated that China-based firms, specifically naming DeepSeek, Moonshot AI, and MiniMax, were involved in wide-scale distillation attacks on their models earlier this year.
menu_book
wikipedia
NEUTRAL
— The roots of the development of artificial intelligence in the People's Republic of China started in the late 1970s following Deng Xiaoping's economic reforms emphasizing science and technology as the…
https://en.wikipedia.org/wiki/Artificial_intelligence_indust…
https://en.wikipedia.org/wiki/Artificial_intelligence_indust…
menu_book
wikipedia
NEUTRAL
— A large language model (LLM) is a type of machine learning model designed for natural language processing tasks such as language generation. LLMs are language models with many parameters, and are trai…
https://en.wikipedia.org/wiki/List_of_large_language_models
https://en.wikipedia.org/wiki/List_of_large_language_models
menu_book
wikipedia
NEUTRAL
— Moonshot AI (Moonshot; Chinese: 月之暗面; pinyin: Yuè Zhī Ànmiàn; lit. 'Dark Side of the Moon') is an artificial intelligence (AI) company based in Beijing, China. It has been dubbed one of China's "AI Ti…
https://en.wikipedia.org/wiki/Moonshot_AI
https://en.wikipedia.org/wiki/Moonshot_AI
+ 3 more evidence sources
“In 2024, the Justice Department indicted a former Google software engineer for stealing AI trade secrets and sharing them with two Chinese companies.”
CORROBORATED
Multiple web search results confirm that the Department of Justice indicted a former Google software engineer (Linwei Ding) for stealing AI trade secrets and sharing them with two Chinese companies.
menu_book
wikipedia
NEUTRAL
— Generative artificial intelligence, commonly known as generative AI or GenAI, is a subfield of artificial intelligence that uses generative models to generate text, images, videos, audio, software cod…
https://en.wikipedia.org/wiki/Generative_AI
https://en.wikipedia.org/wiki/Generative_AI
menu_book
wikipedia
NEUTRAL
— Google LLC ( , GOO-gəl) is an American multinational technology corporation focused on information technology, online advertising, search engine technology, email, cloud computing, software, quantum c…
https://en.wikipedia.org/wiki/Google
https://en.wikipedia.org/wiki/Google
menu_book
wikipedia
NEUTRAL
— Google Search (also known simply as Google or google.com) is a search engine operated by Google. It allows users to search for information on the Web by entering keywords or phrases. Google Search use…
https://en.wikipedia.org/wiki/Google_Search
https://en.wikipedia.org/wiki/Google_Search
+ 3 more evidence sources
“Kratsios argued the abilities of these distilled models may not hold up over time.”
SINGLE SOURCE
The web search results mention Kratsios discussing the limitations of AI models, but the specific claim that he argued the abilities of distilled models 'may not hold up over time' is not independently corroborated by multiple sources.
travel_explore
web search
NEUTRAL
— Michael John Kotsakas Kratsios is an American business executive and government official, who has served as the 13th director of the White House Office of Science and Technology Policy and the science…
https://en.wikipedia.org/wiki/Michael_Kratsios
https://en.wikipedia.org/wiki/Michael_Kratsios
travel_explore
web search
NEUTRAL
— Kratsios, director of the White House Office of Science and Technology Policy, made the remark during a policy address at the Endless Frontiers Retreat in Austin, Texas on April 14. "Our technologies …
https://www.newsweek.com/white-house-says-tech-can-manipulat…
https://www.newsweek.com/white-house-says-tech-can-manipulat…
travel_explore
web search
NEUTRAL
— Echoing President Trump’s agenda, Kratsios declared it is time to reclaim the pioneering spirit that once made the United States the leader in nuclear energy, space exploration, and the Internet.
https://insiderpaper.com/kratsios-our-technologies-permit-us…
https://insiderpaper.com/kratsios-our-technologies-permit-us…
“Kratsios said the Trump administration plans to share intelligence with U.S. AI companies on these campaigns, including the tactics they used, and help the private sector develop defenses.”
SINGLE SOURCE
The claim that Kratsios stated the Trump administration plans to share intelligence with U.S. AI companies on these campaigns and tactics to help develop defenses is mentioned in the context of the overall action plan, but this specific detail is not independently corroborated by multiple sources.
menu_book
wikipedia
NEUTRAL
— The Department of Government Efficiency (DOGE) is an initiative by the second Trump administration in the United States. It was first suggested to Donald Trump by Elon Musk in 2024, and was officially…
https://en.wikipedia.org/wiki/Department_of_Government_Effic…
https://en.wikipedia.org/wiki/Department_of_Government_Effic…
menu_book
wikipedia
NEUTRAL
— Michael John Kotsakas Kratsios (born November 7, 1986) is an American business executive and government official, who has served as the 13th director of the White House Office of Science and Technolog…
https://en.wikipedia.org/wiki/Michael_Kratsios
https://en.wikipedia.org/wiki/Michael_Kratsios
menu_book
wikipedia
NEUTRAL
— Scale AI, Inc. is an American artificial intelligence infrastructure and software company based in San Francisco, California. Originally focused on data annotation, the company also offers RLHF servic…
https://en.wikipedia.org/wiki/Scale_AI
https://en.wikipedia.org/wiki/Scale_AI
info
Disclaimer: This analysis is generated by AI and should be used as a starting point for critical thinking, not as definitive truth. Claims are verified against publicly available sources. Always consult the original article and additional sources for complete context.