US Bans DeepSeek AI From Federal Civilian Agencies Over Security Concerns
The Department of Homeland Security has mandated the removal of Chinese artificial intelligence company DeepSeek from all federal civilian agency systems, marking the latest escalation in Washington's efforts to counter perceived technological threats from Beijing.
Homeland Security Secretary Markwayne Mullin signed the directive on July 24, 2026, though the order was only publicly disclosed on October 20, 2026. The restriction requires civilian federal agencies including the Treasury, State, and Justice departments to eliminate DeepSeek-related technology from their information systems and bars future procurement of the company's products and services. The ban also extends to third-party applications developed using DeepSeek's technology, though it does not apply to the Department of Defense.
Founded in 2023 by Liang Wenfeng, a hedge fund entrepreneur who previously co-founded High-Flyer, one of China's leading quantitative trading firms, DeepSeek gained international attention in January 2025 when its R1 AI model topped Apple's App Store charts in both China and the United States, surpassing ChatGPT in downloads. The company claimed to have developed the model for approximately $5.6 million in training costs, a fraction of the multibillion-dollar budgets typically employed by American tech giants such as OpenAI and Google.
Security Vulnerabilities Identified
According to the DHS order, DeepSeek's AI models present multiple threat vectors that could compromise sensitive government information. The department warned that using the technology could result in the transfer of classified data concerning American citizens and government programmes to Chinese authorities.
Independent security assessments have identified specific technical weaknesses in DeepSeek's systems. NowSecure's evaluation found that the DeepSeek iOS application transmits registration and device data over the internet without encryption, bypassing Apple's App Transport Security protections. The department cited open-source research revealing multiple critical vulnerabilities in the company's mobile application.
In September 2026, the National Security Agency, Cybersecurity and Infrastructure Security Agency, and FBI jointly accused DeepSeek and other Chinese AI firms of conducting industrial-scale campaigns to extract training data from American AI models, adding weight to concerns about the company's operations.
International Response
The United States is not alone in restricting DeepSeek's government use. Taiwan banned the AI service from all government agencies in February 2025, with Premier Cho Jung-tai citing information security concerns and risks of sensitive data reaching China. The Czech Republic followed suit in July 2025 after its national cybersecurity watchdog warned that DeepSeek, as a Chinese company, is legally obligated to cooperate with Chinese government authorities.
DeepSeek's technology differs from Western competitors through its open-source approach, making its internal operations publicly accessible and allowing programmers to customize the software. While this transparency enables inspection, security experts note it also creates potential exploitation pathways.
Previous Restrictions and Legislative Action
Congressional action preceded the executive order, with bipartisan legislation titled the No DeepSeek on Government Devices Act introduced in the House of Representatives in February 2025, followed by a Senate version later that month.
Earlier restrictions had already emerged within specific agencies. The Defense Information Systems Agency blocked access to DeepSeek models in January 2025, while Commerce Department bureaus prohibited staff from using the service on government devices in March 2025. Congress subsequently restricted DeepSeek within intelligence community systems in December 2025.
The latest order permits executive agencies to request exemptions by submitting written justifications to DHS within 60 days of the directive's issuance, including potential extensions to compliance deadlines.
DeepSeek and the Chinese Embassy in the United Kingdom did not respond to requests for comment on the ban.
