The US has banned DeepSeek from federal agencies, including apps built on its AI

I didn’t expect a Chinese startup’s chatbot to land in a US federal procurement order, yet here we are.

Homeland Security has banned DeepSeek from federal civilian agencies. Outside software built on its models is also banned. 

What exactly did the US ban?

Homeland Security disclosed the order on Monday, October 19, 2026. The order directs all civilian agencies to remove DeepSeek products from their networks and stop buying them right away. As I mentioned previously, this also applies to outside software built on DeepSeek’s models (via the Federal Register).

I’d watch that second part closely, since many apps quietly run on borrowed AI models. DHS says DeepSeek runs on servers in China, and local law lets officials dig into company systems. This could eventually risk federal data. 

The interesting bit, however, is that the order covers only civilian agencies. So, the Pentagon is exempt, meaning it can still use it. Other exceptions require a written request. The ban arrives as Washington accuses Chinese labs of copying US AI models. 

If you build or provide services to federal customers, I’d audit your stack for DeepSeek now.

How is China responding?

Beijing answered on Tuesday, October 20, 2026. Mao Ning, China’s foreign ministry spokesperson, said China opposes stretching national security too far and turning tech and trade issues into politics. She also promised to defend Chinese businesses’ lawful rights (via China’s foreign ministry). 

Both governments are still talking, though. In September, they agreed to open a line for talking through AI incidents. Their leaders also meet in Shenzhen next month, so I’d expect this ban to come up there. 

DeepSeek hasn’t commented in the reports I’ve read. 

RELATED COVERAGE: