Last month, DeepSeek, an AI start-up based in China, grabbed headlines with claims that its latest large language AI model, DeepSeek-R1, could perform on par with more expensive and market-leading AI models despite allegedly requiring less than $6 million dollars’ worth of computing power from older and less-powerful chips. Although some industry observers have raised doubts about the validity of DeepSeek’s claims, its AI model and AI-powered application piqued the curiosity of many, leading the DeepSeek application to become the most downloaded in the United States in late January. DeepSeek was founded in July 2023 and is owned by High-Flyer, a hedge fund based in Hangzhou, Zhejiang.
The explosive popularity of DeepSeek coupled with its Chinese ownership has unsurprisingly raised data security concerns from U.S. Federal and State officials. These concerns echo many of the same considerations that led to a FAR rule that prohibits telecommunications equipment and services from Huawei and certain other Chinese manufacturers. What is remarkable here is the pace at which officials at different levels of government—including the White House, Congress, federal agencies, and state governments, have taken action in response to DeepSeek and its perceived risks to national security.
Federal Government-Wide Responses
- Bi-Partisan Bill to Ban DeepSeek from Government Devices: On February 7, Representatives Gottheimer (D-NJ-5) and LaHood (R-IL-16) introduced the No DeepSeek on Government Devices Act (HR 1121). Reps. Gottheimer and LaHood, who both serve on the House Permanent Select Committee on Intelligence, each issued public statements pointing to grave and deeply held national security concerns regarding DeepSeek. Rep. Gottheimer has stated that “we have deeply disturbing evidence that [the Chinese Communist Party (“CCP”) is] using DeepSeek to steal the sensitive data of U.S. citizens,” calling DeepSeek “a five-alarm national security fire.” Representative LaHood stated that “[u]nder no circumstances can we allow a CCP company to obtain sensitive government or personal data.”
- While the details of the bill have not yet been unveiled, any future DeepSeek prohibition could be extended by the FAR Council to all federal contractors and may not exempt commercial item contracts under FAR Part 12 or contracts below the simplified acquisition (or even the micro-purchase) threshold, similar to other bans in this sector. Notably, such a prohibition may leave contractors with questions about the expected scope of implementation, including the particular devices that are covered.
- Other Legislative Activity: Several legislative efforts, targeted more broadly at curtailing Chinese access to AI and other technology, may impact DeepSeek as well. On January 29, Senator Hawley (R-MO) introduced a sweeping U.S.-China AI Decoupling Bill that we previously covered here. If passed, this bill could effectively ban the use of DeepSeek’s AI models in the U.S.
- And on February 7, Representative Mark Green (R-TN-7) introduced the China Technology Transfer Control Act (HR 1122). The bill, among other things, directs the President to “control the export or re-export to or transfer in” China of any “covered national interest technology or intellectual property” under U.S. jurisdiction or exported by any U.S. person. Although identical to a bill introduced by Rep. Green in 2023, Green now frames the bill as a response to DeepSeek and allegations that DeepSeek used U.S. chips and technology to develop its AI models.
- National Security Council Review: On January 28, the White House confirmed that a National Security Council review of DeepSeek’s AI model was ongoing, although no further information has yet been released.
Agency & Component-Specific Responses
- Department of Defense Response: Predictably, the Pentagon and Defense Components have been early movers against DeepSeek. The Defense Information Systems Agency (“DISA”) began blocking access to DeepSeek on the Pentagon’s IT networks on January 28, following reports that Department of Defense employees had used DeepSeek on government devices for about two days before the ban.
- Meanwhile, the U.S. Navy issued a memo on January 24 to service members prohibiting the use of DeepSeek’s AI models “in any capacity” because of “potential security and ethical concerns associated with the model’s origin and usage.” They explained to members of the Operational Navy that it is imperative that team members do not use DeepSeek’s AI “for any work-related tasks or personal use” and that they should “refrain from downloading, installing, or using the DeepSeek model in any capacity.”
- NASA Response: On January 31, NASA’s Chief AI Officer issued a memo to agency personnel prohibiting the use of DeepSeek products and services “with NASA’s data and information or on government-issued devices and networks.” The memo noted that NASA personnel are not permitted to use DeepSeek to “share or upload agency data on DeepSeek products or services,” and are not authorized to “access DeepSeek via NASA devices and agency-managed network connections.”
- U.S. House of Representatives Response: The U.S. House of Representatives’ Chief Administrative Officer issued a notice around January 30 to congressional offices prohibiting House staffers from installing DeepSeek on House-issued phones, computers, and tablets or for “official House use,” noting that “threat actors are already exploiting DeepSeek to deliver malicious software and infect devices.”
State Government Responses
- Texas Issues First State-Level Ban: On January 31, Governor Greg Abbott issued a ban on the use of AI applications affiliated with China, including DeepSeek, on state government-issued devices, making Texas the first state to do so. Abbott stated that “Texas will not allow the Chinese Communist Party to infiltrate [Texas’] critical infrastructure through data-harvesting AI.” He further noted that “state agencies and employees responsible for handling critical infrastructure, intellectual property, and personal information must be protected from malicious espionage operations by the Chinese Communist Party.”
- New York and Virginia Governors Follow: On February 10, New York Governor Kathy Hochul announced a statewide ban on the use of DeepSeek’s AI application on state government-managed devices and networks, citing “serious concerns” related to foreign government surveillance and censorship. On February 11, Virginia became the third state to prohibit DeepSeek on government devices with Governor Glenn Youngkin’s signing of Executive Order 46. The EO prohibits (1) any “employee or contract employee of any agency” from downloading or using DeepSeek’s AI on state government-issued devices, (2) any person or entity contracting with the state from downloading or using DeepSeek’s AI on state-owned or leased equipment, and (3) any person from downloading or using DeepSeek’s AI through state “owned, operated, or maintained wireless networks.”
- State Legislative Activity: While responses to DeepSeek in Texas, New York, and Virginia were led by the executive branch, state lawmakers took the lead in Georgia and Kansas by introducing bills, SB 104 and HB 2313, respectively, that would ban the use of DeepSeek and other Chinese AI models from use on state government-issued devices. These bills remain under consideration.
We anticipate additional legislative and executive agency action regarding the use of DeepSeek by government agencies and their industry partners in the coming weeks and months. This is particularly true given the apparent agreement between key agencies and Congress on the potential risks of this technology. What is clear already is that any use of DeepSeek in connection with U.S. public sector projects will be subject to significant scrutiny.