DHS Orders Federal Agencies to Remove Chinese AI Model DeepSeek

The Department of Homeland Security (DHS) has mandated that federal civilian agencies remove artificial intelligence products developed by China’s DeepSeek from their systems. This directive marks the latest action against the Chinese AI model, which has seen a significant rise in popularity.

The DHS order requires all civilian agency IT systems to purge any products affiliated with DeepSeek, including third-party services that utilize the company's technology. Furthermore, the order prohibits any future procurement of DeepSeek’s products and services.

According to the DHS notice, DeepSeek’s models pose "multiple threat vectors." The department stated that the use of these products "may result in the transfer of sensitive data to the [People’s Republic of China] if used by US government employees or vendors."

The DHS issued this directive under the authority of the Federal Acquisition Supply Chain Security Act of 2018. The decision followed a review of recommendations and evidence presented by the Federal Acquisition Security Council. The notice did not provide specific details about the security concerns or a timeline for agencies to comply.

DeepSeek, founded by Liang Wenfeng, has emerged as one of several Chinese AI models offering more affordable alternatives to prominent US-developed products from companies like Anthropic and OpenAI. While historically trailing behind leading US models in performance, the gap has reportedly narrowed in recent times.

The company gained significant attention early last year with the introduction of its cost-effective models. This development reportedly contributed to a substantial market downturn for major US tech stocks, including Nvidia, Microsoft, and Tesla.

DeepSeek is categorized as an "open-weight" model, meaning its architecture and training parameters can be more easily customized compared to the proprietary, closed models developed by major US tech firms. Open-weight models make their underlying code and training data publicly available, unlike closed models where these elements are kept confidential.

The US government has intensified its scrutiny of China's rapid advancements in AI development and has expressed concerns about alleged efforts by Chinese companies to replicate domestic AI technologies.

Previously, the Commerce Department banned DeepSeek from government devices. The US Navy also issued a directive advising its personnel against using the model due to potential security risks. Similarly, New York State prohibited DeepSeek from its government devices and networks, citing "serious concerns" regarding data privacy and potential censorship risks associated with the application.