House Representatives Josh Gottheimer and Darin LaHood introduced a bill to ban DeepSeek on U.S. government devices over national security concerns and its potential impact on U.S. chipmakers.
On February 7th, House Representatives Josh Gottheimer, D-N.J., and Darin Lahood, R-Ill., introduced the No DeepSeek on Government Devices Act (H.R.1121). This bill seeks to “prohibit the use of DeepSeek by the executive agencies, and for other purposes.”
DeepSeek is a China-based artificial Intelligence (AI) startup that sent shockwaves through Silicon Valley after it unveiled its latest Large-Language Model (LLM), DeepSeek R1, in January of this year. DeepSeek claims that this model is on par with the capabilities of OpenAI's popular LLM, ChatGPT. However, DeepSeek also claims that it was able to develop the R1 model at a fraction of the cost. Just like ChatGPT, DeepSeek can be used as a tool to aid in tasks such as writing emails, answering questions, and coding.
Although the official bill text has not been released yet, we can infer that the DeepSeek Bill would ban the use of DeepSeek by employees and members of the Executive Office of the President as well as all 15 Cabinet Departments. However, the bill would not ban DeepSeek for all Americans.
Firstly, lawmakers and global leaders alike have raised national security concerns over the data security of DeepSeek. Because DeepSeek stores its data on China-based servers, western nations including Italy, Australia, and now the United States have moved to limit its use. Australia has already banned DeepSeek on all government devices and systems. The DeepSeek Bill is indicative of ongoing efforts by the United States government to counter Chinese influence and prevent the acquisition of American data. Similar concerns led to prior legislation, such as the controversial TikTok ban, which stemmed from fears over Chinese access to Americans’ data.
Secondly, the DeepSeek bill is likely a move by the United States government to support US-based chip manufacturers. “DeepSeek has proven that cutting-edge AI models can be developed with limited compute resources,” says Wei Sun, the principal AI analyst at Counterpoint Research. The compute resources that Sun is referring to are primarily Graphics Processing Unit chips, or GPUs. GPUs are the preferred hardware for training AI models–including ChatGPT and DeepSeek–to make them “smarter.” Thus, the release of the R1 model caused US-based GPU manufacturers, such as Nvidia, to plunge in stock value.
Since its introduction, the DeepSeek bill has been referred to the House Committee on Oversight and Government Reform. If passed, it would further restrict the use of foreign AI tools in U.S. government operations, reinforcing ongoing concerns over data privacy and technological independence.
