The U.S. government is mulling a crackdown on Chinese artificial intelligence company DeepSeek, while simultaneously turning up the heat on domestic chipmaker Nvidia amid growing fears of China gaining a technological edge in AI.
Three people familiar with the discussions told The New York Times that the Trump administration is considering penalties to block DeepSeek’s access to U.S. technology and possibly banning access to its services for Americans.
DeepSeek shocked the AI industry in January with the release of DeepSeek-V3, a powerful open-source chatbot reportedly trained for just $6 million, a fraction of the cost typically incurred by U.S. competitors like OpenAI.
Its performance and affordability raised alarm bells in Washington, where officials fear China may be accelerating its AI capabilities to outpace the U.S.
A report published by the U.S. Select Committee on the Chinese Communist Party on April 16 accused DeepSeek of covertly funneling American user data to the CCP, manipulating information to align with CCP propaganda, and training its model using material unlawfully obtained from U.S. AI models.
It also indicated DeepSeek may have trained its model using roughly 60,000 Nvidia chips, including 20,000 restricted under export controls, raising the possibility that U.S. technology found its way into the hands of Chinese developers via unauthorized channels.
It comes as trade tensions between the U.S. and China escalate. The U.S. has imposed tariffs on Chinese goods of up to 145% and tightened export controls on advanced technologies, particularly in artificial intelligence and semiconductors, in a bid to curb Beijing’s rise.
In response, China has raised tariffs on American imports to as high as 125%, deepening the economic rift between the two powers.
In February, Singaporean authorities arrested three people accused of exporting advanced Nvidia chips, which domestic media have claimed were done on behalf of DeepSeek.
The committee is also investigating Nvidia’s sales across Asia, probing whether the company supplied DeepSeek with critical hardware in violation of U.S. rules.
“We now know this tool exploited U.S. AI models and reportedly used advanced Nvidia chips that should never have ended up in CCP hands,” said Committee Chairman John Moolenaar (R-MI) in a statement. “That’s why we’re sending a letter to Nvidia to demand answers. American innovation should never be the engine of our adversaries’ ambitions.”
More details needed
The committee has asked for details on every customer in 11 Asian countries who bought more than 499 AI chips since 2020, including buyers in Malaysia and Singapore, as well as any communications between Nvidia and DeepSeek.
In response, a spokesperson for Nvidia said the company followed the U.S. government’s instructions on where it can buy and sell products to the letter.
“The technology industry supports America when it exports to well-known companies worldwide — if the government felt otherwise, it would instruct us,” Nvidia said.
It added that its reported Singapore revenue indicates the billing address, often for subsidiaries of our U.S. customers. “The associated products are shipped to other locations, including the United States and Taiwan, not to China.”
DeepSeek’s rise has also triggered regulatory scrutiny elsewhere. In January, Italy’s data protection authority ordered the company to block access to its chatbot over privacy concerns, accusing DeepSeek of failing to meet European data standards.
France, South Korea, and Ireland have since signaled their own investigations into the firm’s data collection practices.
DeepSeek did not immediately respond to a request for comment.
Edited by Sebastian Sinclair
Generally Intelligent Newsletter
A weekly AI journey narrated by Gen, a generative AI model.
