The US Department of Homeland Security has issued an order prohibiting federal civilian agencies from utilizing AI technology developed by Chinese lab DeepSeek. According to a Wall Street Journal report, agencies are required to remove existing DeepSeek software from their IT systems and are banned from purchasing future products, extending to third-party services built using DeepSeek implementations.

The directive is rooted in a 2018 US supply chain security law governing federal procurement. However, the order did not specify a deadline for removal, nor did it detail specific security vulnerabilities or clarify how the mandate affects open-weight model implementations and intermediary AI API gateways like OpenRouter.

This decision follows broader US government discussions over restricting open-weight Chinese AI models, such as Moonshot AI’s Kimi K3. While tech companies like Nvidia, Microsoft, Google, and Meta previously pushed back against federal export and usage limits, lobbying efforts from domestic AI leaders like OpenAI and Anthropic favored stricter controls on foreign software.

Why it matters

  • Federal bans on open-weight foreign models alter regulatory risks for enterprise wrapper products and API gateways using open models.

  • Heightened US-China AI competition may accelerate trade restrictions and compliance burdens for US-based developers leveraging foreign open-source code.

Source: the-decoder.com