OpenAI Accuses Chinese Rival DeepSeek of Distilling US AI Models

OpenAI says DeepSeek is using sophisticated methods to extract results from leading US AI models to train its own chatbot.

Published on Feb. 12, 2026

Got story updates? Submit your updates here. ›

OpenAI has warned US lawmakers that its Chinese rival DeepSeek is using unfair and increasingly sophisticated methods, including distillation techniques, to extract results from leading US AI models in order to train the next generation of its breakthrough R1 chatbot. OpenAI says it has detected "new, obfuscated methods" designed to evade its defenses against misuse of its models' output.

Why it matters

This practice of distillation, which is largely tied to China and occasionally Russia, poses a business threat to American AI companies like OpenAI and Anthropic that have invested billions in AI infrastructure and charge subscription fees for their premium services. It also raises national security concerns, as the copied capabilities may lack the safeguards of the original models, enabling more widespread misuse in high-risk areas.

The details

OpenAI says DeepSeek has used distillation techniques as part of "ongoing efforts to free-ride on the capabilities developed by OpenAI and other US frontier labs." The company began raising concerns about this practice shortly after the R1 model's release last year, when it opened a probe with partner Microsoft into whether DeepSeek had obtained its data in an unauthorized manner. OpenAI says it has detected "new, obfuscated methods" designed to evade its defenses against misuse of its models' output.

  • OpenAI sent the memo to the House Select Committee on China on Thursday, February 12, 2026.
  • OpenAI began privately raising concerns about DeepSeek's distillation practices shortly after the R1 model's release last year.

The players

OpenAI

An American artificial intelligence research company that has invested billions of dollars in AI infrastructure and charges subscription fees for its premium services.

DeepSeek

A Chinese rival to OpenAI that has developed an advanced chatbot model called R1.

Microsoft Corp.

A partner of OpenAI that helped the company open a probe into whether DeepSeek had obtained its data in an unauthorized manner.

John Moolenaar

The Republican chair of the House Select Committee on China.

David Sacks

The White House AI Czar who has previously warned about Chinese distillation tactics.

Got photos? Submit your photos here. ›

What they’re saying

“This is part of the CCP's playbook: steal, copy, and kill.”

— John Moolenaar, Republican chair of the House Select Committee on China (Bloomberg)

“Chinese companies will continue to distill and exploit American AI models to their advantage, just like when they ripped off OpenAI to build DeepSeek.”

— John Moolenaar, Republican chair of the House Select Committee on China (Bloomberg)

“DeepSeek was 'squeezing more juice' out of older chips while also citing 'substantial evidence that what DeepSeek did here is they distilled the knowledge out of OpenAI's models.'”

— David Sacks, White House AI Czar (Fox News)

What’s next

The House Select Committee on China is expected to further investigate OpenAI's allegations against DeepSeek in the coming weeks.

The takeaway

This case highlights the growing tensions between the US and China in the race for AI supremacy, with concerns that Chinese companies like DeepSeek are using unfair tactics to gain an edge over their American counterparts. It underscores the need for stronger safeguards and regulations to protect US AI innovations from being exploited.