US lawmakers in Washington DC have this week moved to enact a national ban on the use of DeepSeek, the breakout Chinese generative artificial intelligence (GenAI) tool that sprang to prominence and wiped billions off the value of US tech companies at the end of January.
The No DeepSeek on Government Devices Act is a bipartisan piece of legislation introduced by Democratic congressman Josh Gottheimer and his Republican counterpart, Darin LaHood, who represent districts in the states of New Jersey and Illinois respectively.
The legislation will seek to ban the use and download of DeepSeek’s AI software on government devices. Several other countries have already taken such steps, including the Australian government, which blocked access to DeepSeek on all government devices on national security grounds, and Taiwan. The Italian data protection authority has announced limitations on the processing of Italian users’ data by DeepSeek, and other countries are also considering action.
In the US itself, several bodies have already moved to ban the application, including the state of Texas, which is now restricting its use on state-owned devices, and the US Navy.
Similar to the controversial TikTok ban – currently on hold for 75 days following an executive order signed by President Trump, the US’s attempts to restrict the use of DeepSeek reflect the Western bloc’s long-held concerns over the ability of the Chinese government to co-opt any user data at will from technology organisations.
“The national security threat that DeepSeek – a CCP [Chinese Communist Party]-affiliated company – poses to the United States is alarming,” said LaHood. “DeepSeek’s generative AI program acquires the data of US users and stores the information for unidentified use by the CCP.
“Under no circumstances can we allow a CCP company to obtain sensitive government or personal data. This common-sense, bipartisan piece of legislation will ban the app from federal workers’ phones while closing backdoor operations the company seeks to exploit for access. It is critical that Congress safeguard Americans’ data and continue to ensure American leadership in AI.”
Gottheimer added: “The Chinese Communist Party has made it abundantly clear that it will exploit any tool at its disposal to undermine our national security, spew harmful disinformation and collect data on Americans. Now, we have deeply disturbing evidence that they are using DeepSeek to steal the sensitive data of US citizens.
“This is a five alarm national security fire. We must get to the bottom of DeepSeek’s malign activities. We simply can’t risk the CCP infiltrating the devices of our government officials and jeopardising our national security. We’ve seen China’s playbook before with TikTok, and we cannot allow it to happen again.”
Meanwhile, a separate bill – the Decoupling America’s Artificial Intelligence Capabilities from China Act – introduced by Republican senator Josh Hawley, who represents Missouri and is often outspoken on tech and privacy issues in the US, seeks to penalise the importation of technology or intellectual property developed in China, accompanied by penalties including up to 20 years in prison, and fines of up to $100m for organisations that violate it. Hawley’s bill does not explicitly mention DeepSeek.
“There has been a significant level of nervousness around the use of non-allied technology in government and military settings going back many years. Huawei is just one example,” said Mel Morris, CEO of AI research engine Corpora.ai.
“One could argue that this is just a prudent measure to ensure that devices cannot be compromised by a potential adversary. In the event of a conflict, there are no rules, so whatever assurance or confidence levels might exist would likely go out of the window. I suppose the old adage ‘all’s fair…’. On the flip side, running an air-gapped service using DeepSeek, many would argue, could be made to be safe and considered so.”
Thinking beyond China
Ilia Kolochenko, ImmuniWeb CEO and BCS fellow, said that even though the risks stemming from the use of DeepSeek may be reasonable and justified, politicians risked missing the forest for the trees and should extend their thinking beyond China.
“Numerous other GenAI vendors from different countries – as well as global SaaS platforms, which are now rapidly integrating GenAI capabilities – oftentimes without properly assessing the related risks – have similar or even bigger problems,” he said.
“Whilst DeepSeek’s risks should certainly not be discounted or underestimated, we should remember the fundamental risks and problems of all other GenAI vendors. Many of them unwarrantedly scrapped proprietary and copyrighted content from the internet to train their powerful LLMs – without ever asking for permission from content creators or copyright owners – now vigorously denying any wrongdoing under varying untenable pretexts.
“The unfolding DeepSeek incident shall not be exploited as a ‘convenient’ reason to suddenly forget about serious violations and AI-related risks posed by other GenAI vendors,” said Kolochenko.