The US Department of Homeland Security has ordered civilian agencies to remove DeepSeek products and services built on its models from government IT systems, The Economic Times reports. The order also bars agencies from acquiring them, making this a concrete change to federal technology policy, not just another warning about AI security.
DeepSeek Watch analysis
What happened
The Economic Times says the department issued a notice on Monday. It reports that DHS cited possible transfers of sensitive data to China, potential intelligence gathering and security vulnerabilities, including malware. Those are concerns attributed to DHS, not independently established findings in the report.
The article says a July memo, made public on Monday, also told federal agencies to monitor traffic into and out of government websites for risks linked to DeepSeek. It notes that some agencies, including the Navy and Department of Commerce, had already barred DeepSeek on government devices.
Why it matters
The reported order reaches beyond restricting staff use on individual devices: it also covers procurement and third-party services built using DeepSeek. That gives agencies a wider rule to apply when buying software, although the report does not detail how the notice defines covered services or how agencies must implement it.
The distinction between a government restriction and proof of the underlying security allegations matters. DHS’s concerns are consequential, but they should not be mistaken for independently demonstrated evidence that DeepSeek transferred data or contains malware.
Our read
This is a significant public-sector decision about where AI tools may be used and bought. The useful next step is clarity: agencies and suppliers need to know what counts as a DeepSeek-based service, while the public needs the evidence behind the security claims. A ban can set policy quickly; explaining its scope and basis is the less glamorous work that makes it intelligible.
What to watch
- Whether DHS publishes the notice and gives agencies a deadline or implementation guidance.
- How the government defines third-party services developed using DeepSeek.
- Whether DHS provides supporting evidence for its stated security concerns.
- Whether other agencies extend existing restrictions or adopt the reported order.
Discussion spark: Should governments restrict an AI tool on the basis of stated security risks before making the supporting evidence public, or should disclosure come first?
Sources and evidence
- Source update (19 October 2026, 18:29 UTC)
not affiliated with or endorsed by DeepSeek