US Bans Chinese DeepSeek AI From Federal Systems

Overview

The United States has issued a directive ordering all federal civilian agencies to remove Chinese artificial intelligence provider DeepSeek from their technology infrastructure. The mandate, released by the Department of Homeland Security (DHS), prohibits both the procurement of new DeepSeek services and the continued use of existing systems powered by the Beijing-based startup.

This executive action expands upon previous restrictions, effectively creating a comprehensive ban on DeepSeek’s presence within the US government’s civilian IT ecosystem. The decision marks a significant escalation in the technological cold war between Washington and Beijing, citing irreconcilable national security risks associated with the company’s operations.

The DHS Directive and National Security Concerns

The core of the new order focuses on the potential for data exfiltration and surveillance. According to the DHS, DeepSeek’s operations within China subject it to Chinese laws that require companies to cooperate with state intelligence activities if requested.

The government warns that sensitive information handled by federal employees using DeepSeek’s tools could potentially be accessed by Chinese authorities. This risk is deemed unacceptable for agencies managing classified or proprietary government data.

"The order raises serious concerns about DeepSeek’s handling of user data, warning that sensitive information from US government users could potentially be accessible to Chinese authorities under Chinese law."

The directive is notably broad in its scope. It does not only ban DeepSeek’s proprietary products but also covers third-party software and services built using DeepSeek technology. This means that any vendor offering a solution based on DeepSeek’s underlying models must either replace that component or seek a specific waiver from the government.

While the ban is strict, the DHS acknowledged that exceptions may be granted through a formal request process, allowing agencies to demonstrate that specific uses do not pose a security threat. However, the burden of proof rests heavily on the procuring entities.

Broader Geopolitical Context

This move is the latest chapter in an intensifying rivalry over artificial intelligence supremacy. DeepSeek has disrupted the global AI landscape since the release of its R1 reasoning model in January 2025. The model challenged dominant US rivals like OpenAI and Meta by delivering high-performance results at a fraction of the computational cost.

The financial impact of DeepSeek’s innovations was immediate and severe, triggering a sharp sell-off in Nvidia shares as investors recalibrated their expectations for chip demand. In response, Chinese officials have pushed back against the US actions.

Mao Ning, spokesperson for China’s Foreign Ministry, stated that Beijing opposes the "overstretching" of national security concepts to politicize technology and trade. She emphasized that China would resolutely safeguard the legitimate rights and interests of its enterprises.

The US government had already taken preliminary steps to limit DeepSeek’s access to sensitive sectors. A law passed in December 2025 specifically targeted the use of DeepSeek in intelligence community systems. The new order extends those restrictions to the broader civilian federal workforce, closing what some analysts viewed as a loophole.

Allegations of Model Distillation

Beyond data security, the US has accused DeepSeek of engaging in model distillation practices to extract intellectual property from American systems. A September report alleged that DeepSeek, along with five other Chinese AI companies including Alibaba and Moonshot AI, used large-scale distillation to mimic the outputs of Western AI models.

Model distillation involves training one AI system using the outputs generated by another. US tech giants argue that this technique allows Chinese firms to bypass the massive investment required to train foundational models from scratch.

OpenAI and Anthropic have independently raised alarms regarding these activities. OpenAI reported to US lawmakers that it had observed suspected attempts by DeepSeek to distill its proprietary models. Similarly, Anthropic stated that it had identified unauthorized efforts by DeepSeek and other Chinese labs to collect outputs from its Claude language models.

These allegations have further strained diplomatic and technical relations, reinforcing the US stance that the risks posed by DeepSeek extend beyond mere data privacy into the realm of intellectual property theft.

China’s Countermeasures and Global Scrutiny

In a potential retaliatory move, reports indicate that China is considering export restrictions on 23 AI-related technologies. These measures, first highlighted by Bloomberg, could include controls on advanced chip designs and certain high-tier AI models. Chinese regulators aim to address security concerns and prevent the misuse of critical technologies abroad.

DeepSeek has faced scrutiny from governments and regulators worldwide since its rise to prominence. The company’s rapid ascent has forced other nations to reevaluate their own policies regarding Chinese AI adoption. As the US solidifies its ban, other allies may follow suit, potentially fragmenting the global AI market along geopolitical lines.

For federal agencies, the coming months will involve a logistical challenge of auditing their software stacks, identifying DeepSeek dependencies, and migrating to compliant alternatives. The outcome of this transition will likely serve as a barometer for the broader decoupling of US and Chinese technological ecosystems.