OpenAI has claimed that Chinese companies are attempting to develop their own models by using U.S. artificial intelligence technologies and announced that it will take serious measures to prevent this. The company intends to collaborate with the U.S. government to investigate potential national security risks. According to Reuters, this statement came after the White House suggested that Chinese-origin DeepSeek could be a potential threat to the U.S.
David Sacks, the White House official responsible for AI and cryptocurrency policy, stated in an interview with Fox News that Chinese companies could learn from OpenAI models using a method called "distillation."
Founded in 2023 in China, DeepSeek works on open-source AI models. In 2024, the startup launched the initial version of its R1 model with reasoning capabilities, followed by the full version in January 2025. The company claims that the R1 model performs equally well or better than OpenAI's o1 model in programming, mathematics, and logic tasks.
DeepSeek has generated significant interest. According to App Figures, the app had been downloaded 1.6 million times by January 25. However, the company temporarily suspended new user registrations due to a cyberattack.
DeepSeek's success has negatively affected the stock prices of tech companies such as Nvidia and AMD. Investors are concerned that DeepSeek's efficient model training and operation methods could reduce the demand for chips. Nevertheless, some experts believe that the market panic is exaggerated.
On January 28, after the DeepSeek app topped the U.S. App Store, it also became the most downloaded app in Russia. Currently, DeepSeek ranks seventh among free services on Russian Google Play.
