US bars DeepSeek from federal civilian agencies’ systems
The US Department of Homeland Security is also banning services that process DeepSeek models.
Listen to this article
There is no audio version yet. Request one and an AI voice will read the article aloud.
Read by an AI voice.
The United States is ordering federal civilian agencies to remove products from Chinese AI developer DeepSeek from their computer systems. New government purchases of DeepSeek products and services that use the models are also excluded.
The decision comes from the US Department of Homeland Security. An official notice states that products from Hangzhou DeepSeek Artificial Intelligence and affiliated companies may no longer be procured by federal civilian agencies.
Existing DeepSeek products must also be removed from those agencies’ information technology systems. According to the notice, the measure also applies to third-party services that use DeepSeek implementations. The decision therefore targets not only the chatbot or an individual app, but also software in which the model is embedded.
The department points to risks to the security of government systems. The official text names no public technical incident as the reason for the decision. Reuters reported that the US government sees the measure as a response to possible risks surrounding DeepSeek’s activities in China.
The move fits into a broader US approach in which technology from China is assessed for possible implications for national security and data protection. The United States has previously also taken measures against other Chinese technology companies and apps, although the legal grounds differ from case to case.
For Dutch organisations, the US decision does not directly change the rules. It may, however, affect international software suppliers that hold US government contracts. They will have to establish whether their products use DeepSeek models, even when this is not visible to the end user.
The measure in itself says nothing about the performance or safety of each individual DeepSeek model in every application. The US government cites security risks as the reason for the measure.
The notice will be published in the Federal Register on 20 October. Until then, the version published in advance is the available official source.
Also show the region and a close-up
One story, several perspectives
What is established
- The US Department of Homeland Security has imposed a ban on the use and procurement of DeepSeek in federal civilian systems.
- The measure also applies to third-party services that use DeepSeek technology.
- The US government cites security risks as its justification.
- No public evidence has been released that DeepSeek misused data in this specific case.
Left
Arguments Governments should not impose broad exclusion measures without concrete, verifiable evidence. Open standards and independent audits are preferable to nationality as the main criterion.
Values Non-discrimination, transparency and verifiable facts.
Consequences A ban could restrict innovation and affordable AI, and could reinforce international digital fragmentation.
Centre
Arguments Government agencies may impose stricter requirements as a precaution on software that processes sensitive data. The preference should be for risk assessment, transparency about model use and proportionate measures.
Values Digital security, legal certainty and practical feasibility.
Consequences Targeted exclusion of high-risk applications can provide protection without automatically banning all Chinese technology.
Right
Arguments The state must retain maximum control over data, suppliers and dependencies in sensitive systems. Where there are doubts about foreign influence, exclusion is defensible, particularly in federal infrastructure.
Values National security, sovereignty and resilience.
Consequences A strict ban could reduce risks, but could also cause trade conflicts and higher costs for government IT.
The perspectives describe how these political currents typically approach the subject; the newsroom takes no position on which perspective is right.
Fact-check Approved · Nour Haddad — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The scope of the US decision was checked directly against the official notice and confirmed by Reuters and AFP. The text avoids the unsubstantiated claim that DeepSeek actually stole data.
- confirmed US federal civilian agencies must remove DeepSeek products. — Stated in the official DHS notice. source
- confirmed Third-party services that use DeepSeek implementations are also covered by the measure. — Explicitly mentioned in the summary of the decision. source
- confirmed The US government justifies the measure on the basis of security risks. — The measure and reporting cite national security and cyber risks as the reason. source
Editor's note
The removal and procurement restrictions are set out in the official US notice. The stated security concerns are presented as the government’s justification; no specific public incident proving this measure was found.Sources
More on this in Dutch media
- NOS — „deepseek kunstmatige intelligentie”
- Het Parool — „deepseek kunstmatige intelligentie”
- NRC — „deepseek kunstmatige intelligentie”