AI Theft on a Massive Scale: US Accuses China of Copying Top Models
The United States government has fired a major accusation at China, claiming that foreign entities primarily based there have been running "industrial-scale" campaigns to copy frontier American AI models. According to a memo reported by CNN, White House director of the office of science and technology policy Michael Kratsios made the allegations on Thursday, putting a spotlight on one of the most alarming developments in the global AI race.
What Exactly Is the White House Saying?
Michael Kratsios, who leads the White House office of science and technology policy, published a memo on Thursday laying out the accusations in stark terms. He stated that foreign entities, primarily based in China, have been systematically carrying out large-scale efforts to extract capabilities from American AI models. The memo describes these efforts as coordinated campaigns that exploit American expertise and innovation at an unprecedented scale.
Tens of Thousands of Fake Accounts Involved
One of the most striking details in the Kratsios memo is the sheer scale of the operation. According to the memo, these campaigns use tens of thousands of surrogate accounts to go undetected. On top of that, complex technical tools are reportedly being deployed to expose proprietary information belonging to American AI companies. This is not a case of casual copying. It is an organized, systematic effort built to evade detection while quietly draining American AI knowledge.
The Technique at the Center: AI Distillation
At the heart of these accusations is a process known as "distillation." This is a technique used to transfer knowledge from a large, powerful AI model to a smaller one that can operate more cheaply and efficiently. Distillation itself is a widely accepted and legitimate method used across the AI industry for training models. However, companies like Anthropic and OpenAI have previously raised alarms that distillation has been used to unfairly mimic their models' capabilities without authorization.
DeepSeek: The Chinese AI Startup at the Center of It All
DeepSeek, the Chinese AI startup that sent shockwaves through Wall Street last year, sits at the center of these accusations. OpenAI warned US lawmakers in February that DeepSeek had been attempting to replicate the performance of ChatGPT and other American frontier AI models through distillation. The sudden rise of DeepSeek had already rattled the tech industry. These new accusations add a much more serious dimension to its story, raising questions about how much of its capability was built on the back of American innovation. China's broader ambitions in AI are no secret, and Baidu's aggressive AI integration push is another example of how Chinese tech giants are racing to close the gap with their American counterparts.
Anthropic's Own Warning from February
Anthropic, the AI safety company behind the Claude family of models, also went public with similar concerns back in February. The company stated that it had identified what it described as industrial-scale campaigns by DeepSeek and two other AI labs to illicitly extract Claude's capabilities to improve their own models. Anthropic published this finding in a blog post, making clear that the problem was not hypothetical. The company had actually detected and documented these distillation attacks against its own systems.
Why Distilled Models Are a Security Risk
Beyond the issue of intellectual property theft, distilled AI models carry a serious security concern. Anthropic noted in its February blog post that these models lack the safety guardrails built into the original models they were trained on. The Kratsios memo echoes this concern directly, stating that distillation campaigns allow actors to deliberately strip security protocols from the resulting models. He also warned that such models can have their mechanisms for ideological neutrality and truth-seeking removed, making them potentially dangerous tools in the wrong hands.
The Bigger Picture: US vs China in the AI Race
The United States and China have been locked in an intense competition to build the most advanced AI technology. This rivalry has become a major flashpoint in trade tensions between the two superpowers. Companies like Nvidia have already found themselves caught in the middle, facing restrictions on chip exports to China. The AI competition is not just about products and profits. It is increasingly seen as a matter of national security and global technological leadership, and accusations of large-scale AI theft only heighten the stakes further.
China's Tech Ambitions Keep Growing
China has made it crystal clear that it intends to be a dominant force in the technology world, and AI is only one piece of that ambition. The country has been aggressively pushing its tech ecosystem forward across multiple sectors. From AI-powered applications to industrial robotics, China is advancing at a rapid pace. It is worth noting that China now leads the world in robotics, reflecting just how broad and deep its technology ambitions run across industries.
What the Trump Administration Plans to Do
The Trump administration has outlined several concrete steps it plans to take in response to these alleged campaigns. First, it intends to share information with US AI companies about attempts to conduct unauthorized, industrial-scale distillation of their models. Second, it plans to improve coordination between private companies to better defend against these threats. The administration is also exploring measures to hold foreign actors accountable for such campaigns and will work with the private sector to develop best practices for detecting and defending against industrial-scale distillation activities.
Trump's Broader AI Strategy
Cementing the United States as a global leader in AI has been a cornerstone of President Trump's second term. His administration has pushed for federal AI regulation rather than a patchwork of state-level rules, with the goal of speeding up innovation. However, critics have argued that this approach could make it easier for AI companies to evade accountability. Separately, Trump has already moved to tighten controls on the sale of AI chips to China, an effort that reflects the administration's view of AI as a strategic national asset.
No Response from DeepSeek or China's Embassy
As of the time of CNN's report, DeepSeek had not responded to a request for comment on the accusations. CNN also reached out to the Chinese embassy in Washington for a response, but no comment had been received at the time of publication. The silence from both parties leaves the accusations from the White House standing without a direct rebuttal, at least for now.
What This Means for the Future of AI
The White House memo signals that the United States is taking AI theft very seriously and is preparing to treat unauthorized distillation campaigns as a geopolitical and security issue. For AI companies, this is a wake-up call. The methods they use to protect their models from distillation attacks will need to become far more sophisticated. For policymakers, the memo makes clear that the AI race is not just about who can build the smartest model. It is also about who can protect their innovations in the face of aggressive and well-organized attempts to steal them.
Source & AI Information: External links in this article are provided for informational reference to authoritative sources. This content was drafted with the assistance of Artificial Intelligence tools to ensure comprehensive coverage, and subsequently reviewed by a human editor prior to publication.
Posts
0 Comments