DHS Bans DeepSeek AI in Federal IT Citing Supply Chain Risks, Halts Future Procurement

The Department of Homeland Security has issued a directive requiring all US federal civilian agencies to remove artificial intelligence technology developed by DeepSeek from their information technology systems. This order also prohibits the future procurement of any products or services from the company. The move aims to mitigate potential national security risks associated with the vendor’s software and data practices.

Signed by Homeland Security Secretary Markwayne Mullin and dated 24 July, the order was posted on the Federal Register website on Monday. According to Bloomberg, the directive bars DeepSeek from participating in federal contracts. The action applies specifically to civilian agencies and extends beyond direct DeepSeek products to include third-party tools that are built upon DeepSeek’s underlying AI models.

What we know

The order was issued under a 2018 law that allows the government to remove risky suppliers from its buying processes. According to Bloomberg, the government stated that DeepSeek collects significant amounts of user data that could be passed to the Chinese government under Chinese security laws. Reuters reported that the department also cited the risk of the models posing multiple threat vectors, including the potential to hide malware or security flaws.

Some government entities had already restricted the Chinese startup’s tools prior to this federal order. A Pentagon IT agency blocked DeepSeek’s chatbot in January 2025, while the Commerce Department and the Navy had previously banned it on work devices. Several states, including New York, Tennessee, and Virginia, have taken similar steps to restrict access to these tools on official networks.

This directive follows months of political pressure on Chinese AI firms in Washington. In June, the White House delayed blacklisting DeepSeek and over 100 other Chinese firms, despite a panel of officials clearing them for a US trade blacklist. US officials have also accused Chinese AI labs of copying US models through distillation, a technique OpenAI told lawmakers DeepSeek used to build its R1 chatbot in February.

Defender's checklist

• Verify if any internal systems use DeepSeek models or third-party wrappers built on them.
• Review procurement contracts for any clauses involving DeepSeek products or services.
• Check for unauthorized third-party AI tools that may rely on DeepSeek infrastructure.
• Monitor network traffic for data exfiltration patterns linked to known DeepSeek endpoints.

The background in plain language

Organisations often rely on third-party software to handle complex tasks like data analysis or customer support. When these tools process sensitive information, there is a risk that the data could be accessed by foreign entities or malicious actors. Supply chain security laws allow governments to cut ties with vendors whose practices might compromise national security. Teams usually assess whether a vendor’s data handling aligns with local privacy laws and security standards before adoption.

What to do and how to stay safe: DeepSeek

• Audit current software inventories to identify any tools that utilise DeepSeek models or APIs.
• Review data handling policies to ensure sensitive information is not being sent to unapproved external services.
• Monitor official government channels for updates on banned vendors and updated security guidelines.
• Implement strict access controls for AI tools to prevent unauthorised usage on corporate networks.

General security guidance from the TechWire Beat newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Does this ban apply to all federal agencies?

The order specifically applies to US federal civilian agencies, though some military branches had already imposed their own restrictions.

Why is DeepSeek being banned from federal systems?

The DHS cited risks that user data could be shared with the Chinese government and that the models could hide malware or security flaws.

Does the ban cover third-party applications?

Yes, the order covers third-party tools that are built on DeepSeek’s AI models, not just direct products from the company.