OpenAI Says China’s DeepSeek May Have ‘Inappropriately’ Used Its Data to Build Rival Model

The firm has attracted top U.S. and Australian officials’ attention for national security implications.

Screens displaying the logo of DeepSeek, a Chinese artificial intelligence company that develops open-source large language models, and the logo of OpenAI's artificial intelligence chatbot ChatGPT, in Toulouse, France, on Jan. 29, 2025. Lionel Bonaventure/AFP via Getty Images

Chinese artificial intelligence (AI) startup DeepSeek has stunned the market by introducing a chatbot built at a fraction of the cost of U.S. rivals. But questions are growing over whether the firm may have taken a shortcut.

ChatGPT creator OpenAI said on Jan. 29 that DeepSeek might have “inappropriately” used its data.

“We know that groups in the PRC are actively working to use methods, including what’s known as distillation, to try to replicate advanced U.S. AI models,” an OpenAI spokesperson told The Epoch Times, using the acronym for the Chinese regime’s official name, the People’s Republic of China.

“We are aware of and reviewing indications that DeepSeek may have inappropriately distilled our models, and will share information as we know more.”

Distillation is an AI technique in which a developer trains an AI model by siphoning data from a larger one. OpenAI, in its terms of service, states that it does not allow anyone to take data from its system to build competing products. David Sacks, the White House AI czar, suggested DeepSeek has done just that.

“There’s substantial evidence that what DeepSeek did here is, they distilled the knowledge out of OpenAI’s models,” he said in a recent Fox News interview. “I don’t think OpenAI is very happy about this.”

Howard Lutnick, President Donald Trump’s pick for secretary of the Commerce Department, voiced similar skepticism about DeepSeek in his Senate confirmation hearing on Jan. 29.

“I do not believe that DeepSeek was done all above board. That’s nonsense,” he said. “I’m going to be rigorous in our pursuit of restrictions and enforcing those restrictions to keep us in the lead.”

On Jan. 27, DeepSeek overtook OpenAI’s ChatGPT as the top free downloaded app on Apple’s App Store globally, sparking fears among investors.

Founded in 2023, the Chinese AI firm claimed to have built the model in two months with less than $6 million. The Chinese media celebrated the company’s cost-efficiency, calling it the “Pinduoduo of AI,” a reference to the Chinese online retailer known for its ultra-low pricing.

Little is mentioned about the app’s alignment with the ideology of the Chinese Communist Party (CCP).

In testing the AI assistant tool, The Epoch Times found that DeepSeek repeatedly shied away from human rights topics. It refused to answer questions about the Falun Gong Protection Act, a legislative bid in the 118th Congress to counter the CCP’s ongoing persecution of the spiritual group, including forced organ harvesting.

When prompted to describe The Epoch Times, which reports critically on Beijing and is banned in China, the app initially said the publication is “known to publish content critical of the Chinese government and the Communist Party of China” and then quickly deleted the response.

The app said it follows “Chinese laws, regulations, and socialist core values” and is “committed to providing services that are in line with national conditions and cultural backgrounds.”

Chinese officials appear to view the company favorably.

On Jan. 20, DeepSeek’s founder, Liang Wenfeng, was one of nine representatives invited to speak at a seminar chaired by Chinese Premier Li Qiang.

DeepSeek released the open-source model R1 the same day.

Top U.S. and Australian officials have been considering potential national security risks with the app, which New York City-based cybersecurity firm Wiz said had accidentally exposed more than a million lines of sensitive data to the open internet, including chat history and backend details.

DeepSeek said it stores information in servers located in China.

In her first press briefing on Jan. 28, White House press secretary Karoline Leavitt said the National Security Council is looking into the national security implications.

Ed Husic, Australia’s science minister, cautioned on the same day against downloading the Chinese app, saying, “There are a lot of questions that will need to be answered in time on quality, consumer preferences, data, and privacy management.”

White House press secretary Karoline Leavitt holds her first news conference in the James S. Brady Press Briefing Room at the White House on Jan. 28, 2025. Chip Somodevilla/Getty Images

Lawmakers have been calling for steps to restrict Chinese AI growth.

Sen. Josh Hawley (R-Mo.) on Jan. 29 introduced a bill to bar U.S. companies from investing in Chinese AI development or collaborating with the Chinese regime on AI research. The legislation also aims to limit the transfer of relevant technologies to China.

Reps. John Moolenaar (R-Mich.) and Raja Krishnamoorthi (D-Ill.), chairman and ranking member of the House Select Committee on the Chinese Communist Party, respectively, wrote to national security adviser Mike Waltz, urging him to consider putting export controls on semiconductor chips crucial to DeepSeek’s AI system.

“Like any claims from China, Americans should be skeptical of DeepSeek’s self-reported development costs,“ Moolenaar said in a statement. “In this case, the rapid growth and CCP control behind the company raises serious national security concerns and demands immediate action.”

Krishnamoorthi called DeepSeek “deeply alarming,” adding that export controls and AI innovation are “two sides of the same coin.”

“If we want to outcompete the CCP in AI, we must protect our lead, safeguard Americans’ data, and use common sense,” he said.