DeepSeek’s rapid rise is now being matched by sharper scrutiny. After its openly available models and free chat apps drew major attention this week, Trump’s AI and crypto "czar," David Sacks, said the Chinese AI company may have used OpenAI’s models to improve its own.
What David Sacks alleged
In an interview on Fox on Tuesday, Sacks said there is "substantial evidence" that DeepSeek "distilled" knowledge from OpenAI’s AI models. He compared that process to theft.
Sacks did not cite the source of this "evidence." The central allegation, as described in the source article, is that DeepSeek used responses from OpenAI models to train its own systems.
That distinction matters because the concern is not simply that DeepSeek built strong AI models. The concern raised by Sacks is about how those models may have been trained and whether OpenAI’s outputs were used as part of that process.
"I don’t think OpenAI is very happy about this," Sacks said.
The source article does not include a response from DeepSeek or OpenAI. It also does not provide the evidence Sacks referenced, so the claim remains presented as an allegation from Sacks rather than a documented finding in the article.
Why DeepSeek is getting attention now
DeepSeek went viral this week because of its openly available, highly performant models and free chat apps. That combination has made the company a focal point in the AI industry, especially as users and officials look more closely at how capable models are being built and distributed.
The company’s apps also topped the Apple App Store this past weekend. That level of public visibility helps explain why the discussion has quickly moved beyond technical circles and into government review.
The source article presents two tracks of attention happening at once:
Industry attention around DeepSeek’s openly available and highly performant AI models.
Government scrutiny over the implications of DeepSeek’s apps and reported security and ethical concerns.
Taken together, those tracks show why the DeepSeek OpenAI issue has become broader than a dispute about model training. It now touches on access, competition, public adoption and institutional risk.
What "distilled" means in this case
The article describes the alleged process in plain terms: DeepSeek may have used responses from OpenAI models to train its own. Sacks called that "distilled" knowledge.
Based only on the source, the key idea is that a model’s outputs can become training material for another model. If that happened, the debate would center on whether DeepSeek benefited from OpenAI’s work by learning from OpenAI-generated responses.
The article does not provide technical details about how any such process would have worked. It does not name specific OpenAI models, training datasets, systems or methods. It also does not state when any alleged use took place.
That leaves the public discussion with a clear claim but limited supporting detail in the article: Sacks says there is evidence, but he did not identify where it came from.
Government scrutiny is widening
DeepSeek is being looked at with increasing scrutiny by U.S. government officials. According to Reuters, the National Security Council, the U.S. body that advises the president on foreign policy and national security, is reviewing the implications of DeepSeek’s apps.
The U.S. Navy has also reportedly banned the use of DeepSeek’s AI. The stated reason in the source article is "potential security and ethical concerns." The article does not expand on those concerns or describe the scope of the ban beyond that reported action.
Those two developments show that DeepSeek is now being evaluated not only as a technology product but also as a potential policy and security issue. The App Store ranking matters here because widely used apps can quickly become relevant to institutions that must assess data, access and trust risks.
Still, the source article is careful in what it establishes. It reports that officials are reviewing implications, that the Navy has reportedly banned use, and that Sacks made an allegation about distillation. It does not state that any investigation has reached a conclusion.
The bigger question for AI companies
The DeepSeek story highlights a practical issue for the AI sector: as more companies release capable models and chat apps, questions about model training will become harder to separate from questions about competition and trust.
For users, the immediate appeal is straightforward. DeepSeek’s models are described as openly available and highly performant, while its chat apps are free. For officials and rival companies, the concern is different: how those capabilities were achieved and what risks may come with broad adoption.
The source article does not resolve those questions. It does, however, show the pressure points clearly. DeepSeek’s success has made it visible, Sacks’s comments have raised questions about OpenAI model distillation, and U.S. government bodies are now paying attention to the company’s apps.
Until more evidence is made public, the most accurate reading is that DeepSeek is facing serious scrutiny based on allegations and reported government concern, not a publicly documented conclusion in the article.