DHS Mandates Federal Agencies to Purge DeepSeek AI Tools from IT Networks
Homeland Security orders removal of DeepSeek technology and bans future procurement citing supply chain risks and data concerns.
Key points
- The order, dated July 24, covers civilian agencies and third-party tools built on DeepSeek models.
- DHS cites risks of data sharing with Beijing and potential malware hidden in the AI models.
- The ban is issued under a 2018 supply chain security law, barring DeepSeek from federal contracts.
The US Department of Homeland Security has ordered all federal civilian agencies to remove artificial intelligence technology developed by DeepSeek from their information technology systems. This directive also prohibits the future procurement of DeepSeek products and services, aiming to mitigate supply chain risks associated with the Chinese startup.
According to the Wall Street Journal, the order requires agencies to remove DeepSeek tech from their networks. The Department of Homeland Security posted the notice on the Federal Register website on Monday, formalizing the ban. The directive was signed by Homeland Security Secretary Markwayne Mullin on July 24, as reported by Bloomberg.
Scope and Security Rationale
The ban extends beyond direct DeepSeek software to include third-party tools that are built on DeepSeek’s AI models. DHS cited the risk that sensitive government data could be shared with the Chinese government under Chinese security laws, according to Bloomberg. The department stated that DeepSeek collects significant user data, creating a pathway for foreign access.
Reuters reported that DHS also warned the models pose multiple threat vectors. The agency noted that the AI systems may hide malware or contain security flaws that could compromise federal networks. The order was issued under a 2018 supply chain security law that allows the government to exclude risky suppliers from its buying processes.
Affected Agencies and Prior Actions
The mandate applies specifically to civilian federal agencies, barring DeepSeek from participating in federal contracts. Some government entities had already taken steps to limit the startup’s tools before this comprehensive order. In January 2025, a Pentagon IT agency blocked DeepSeek’s chatbot from its systems.
The Commerce Department and the Navy have also banned DeepSeek on work devices, indicating a growing internal consensus on the risk. Similar restrictions were adopted by New York, Tennessee, and Virginia at the state level, according to The Next Web. These prior actions suggest the DHS order codifies existing security postures into a unified federal policy.
Broader Context on Chinese AI Firms
This directive follows months of increased pressure on Chinese AI firms in Washington. In June, the White House decided against blacklisting DeepSeek and more than 100 other Chinese firms, despite a panel of officials recommending their inclusion on a US trade blacklist, Bloomberg noted. This hesitation contrasts with the current decisive action taken by DHS.
US officials have accused Chinese AI labs of copying US models through a technique called distillation. In February, OpenAI told US lawmakers that DeepSeek used this method to build its R1 chatbot, according to Bloomberg. The Chinese embassy in Washington did not immediately respond to requests for comment, and DeepSeek did not comment in either report.
What should you do?
Federal IT administrators must audit their systems for DeepSeek components. They should verify that no third-party applications rely on DeepSeek models. Agencies should update their procurement guidelines to reflect the new ban. Monitoring for new DHS guidance will be essential as the order is implemented.
What to do and how to stay safe: DeepSeek
- Conduct an inventory of all AI tools and third-party integrations to identify any reliance on DeepSeek models.
- Review procurement contracts to ensure no future agreements involve DeepSeek or its derivatives.
- Monitor official DHS channels for updates on implementation timelines and specific compliance requirements.
- Assess data handling practices to ensure no sensitive information has been exposed through previous use of these tools.
Step-by-step guide: Mobile Device Security: How to Protect Data Beyond Passwords
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
Does this ban apply to all US government agencies?
No, the order specifically applies to federal civilian agencies, though the Pentagon and Navy had issued their own bans earlier.
Why is DHS banning DeepSeek technology?
DHS cites risks that sensitive data could be shared with the Chinese government and that the models may hide malware or security flaws.
Are third-party tools affected by this order?
Yes, the ban covers tools from other firms that are built on DeepSeek’s AI models, not just direct DeepSeek products.
