Chinese AI developers are dominating the open-source model landscape, not primarily through alleged data extraction campaigns, but because they embrace a collaborative ecosystem while U.S. companies operate in isolated silos. This structural difference in how the two regions approach AI development may matter more than any single distillation controversy, according to industry observers tracking the intensifying competition between American and Chinese artificial intelligence capabilities.

What Are Chinese AI Companies Actually Accused Of?

On September 9, 2026, the National Security Agency (NSA), Cybersecurity and Infrastructure Security Agency (CISA), and Federal Bureau of Investigation (FBI) issued a joint advisory naming six Chinese AI companies of conducting large-scale campaigns to extract capabilities from U.S. frontier models. The companies named were DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI. According to the agencies, these companies extracted billions of tokens through millions of requests to frontier models including Claude, GPT, Gemini, and Grok since at least late 2024.

The practice at the center of the controversy is called knowledge distillation, a machine learning technique where one model is trained using the outputs of another model. While distillation is a legitimate training method used widely in the industry, the agencies alleged that these companies engaged in "aggressive, malicious and targeted" distillation at industrial scale to obtain restricted proprietary functions without bearing comparable development costs.

The advisory specifically highlighted DeepSeek's alleged organized campaign since late 2024 to obtain reasoning capabilities and specialized optimizations for its R1 and V3 models. Anthropic had previously reported in February 2026 that it identified industrial-scale campaigns by DeepSeek, Moonshot AI, and MiniMax involving more than 16 million exchanges with Claude through approximately 24,000 fraudulent accounts that violated terms of service and regional access restrictions.

How Does China's Collaborative Model Differ From the U.S. Approach?

Beyond the distillation allegations, a more fundamental competitive advantage may lie in how Chinese and American AI developers approach open-source development. Hugging Face CEO Clement Delangue observed in August 2026 that Chinese developers were clearly dominating in open models, attributing China's position to an open collaboration and sharing ecosystem.

"Chinese developers were clearly dominating in open models. U.S. model makers were building in silos and risked falling behind," Delangue noted, highlighting a structural difference in how the two regions organize their AI development efforts.

Clement Delangue, CEO at Hugging Face

This observation suggests that the competitive gap extends beyond any single incident of alleged data extraction. The concentrated structure of the U.S. AI industry, where a handful of large corporations hold most resources, may be creating barriers to the kind of collaborative innovation that characterizes China's approach.

Steps to Understand the Distillation Controversy

• What distillation is: A legitimate machine learning technique where one model learns from another model's outputs, commonly used to create smaller or more efficient versions of larger models.
• How it was allegedly misused: The six Chinese companies allegedly used distillation at industrial scale to extract proprietary capabilities from U.S. frontier models without bearing the full development costs, using third-party services to bypass geographic restrictions and safeguards.
• Why it matters for competition: If successful, distillation campaigns could allow Chinese AI developers to narrow the technology gap while avoiding some of the computing power, electricity, and foundational research costs required to independently develop frontier models.

The agencies assessed that the alleged activity took place "likely with Chinese government awareness," though the advisory did not state that Chinese authorities directed the individual campaigns. Beijing rejected the allegations on September 9, with Chinese Ministry of Foreign Affairs spokeswoman Mao Ning stating that China's AI development came from domestic technological development and international cooperation.

What Do U.S. Officials Say About the Competitive Stakes?

Treasury Secretary Scott Bessent framed the AI competition with China as existential for American technological leadership. Speaking at Breitbart News' "State of the Economy" event in Washington on September 9, Bessent warned that if China pulled ahead in AI, "nothing else matters".

Bessent

"Beating China, there is no day after tomorrow if China wins at this. If they were to pull ahead of us on AI, then nothing else matters," Bessent stated, emphasizing the stakes of the competition.

Scott Bessent, U.S. Treasury Secretary

Bessent tied the competition to the U.S. buildout of data centers and infrastructure needed for AI development. He argued that the United States currently has the lead because of its technology companies, advanced chips, financing, and startup ecosystem, but that development could not be paused. "We can't pause," he said. "You can't, because the Chinese won't pause."

However, Bessent criticized technology companies for failing to engage adequately with communities affected by new data-center construction, giving data-center developers, hyperscalers, and major AI companies a "D-minus" for community outreach.

What Actions Are U.S. Agencies Taking in Response?

The September 9 advisory from NSA, CISA, and FBI urged American AI companies and infrastructure providers to improve detection of suspicious activity and strengthen coordination across model developers, cloud providers, and API providers to counter large-scale distillation campaigns. The agencies recommended that companies monitor for patterns of unusual request volumes, geographic anomalies, and obfuscated access methods that might indicate distillation attempts.

OpenAI separately told the House Select Committee on the Chinese Communist Party in February 2026 that it had observed activity indicating continued attempts by DeepSeek to distill capabilities from OpenAI and other U.S. frontier models, including through increasingly obfuscated methods intended to make the activity harder to detect. This suggests that detection and prevention remain ongoing challenges as techniques become more sophisticated.

The broader context includes a pattern of Chinese technical replication. In December 2024, researchers from Fudan University and the Shanghai AI Laboratory successfully replicated OpenAI's advanced o1 reasoning model, a key step toward replicating frontier AI capabilities. The subsequent release of DeepSeek's open-source reasoning model sent shockwaves through the tech industry, with the model rivaling OpenAI's systems at a fraction of the cost and running on standard hardware.

As the competition intensifies, the question facing U.S. policymakers and technology leaders is whether addressing distillation campaigns alone will be sufficient, or whether structural changes to how American AI companies collaborate and share knowledge will be necessary to maintain technological leadership in the long term.