Homeland Security just BANNED this popular AI from the government

Artificial intelligence has quickly evolved from an interesting novelty to an economic driver and a cybersecurity threat. But while the United States tries to rein in its domestic models, foreign-owned AI could pose an even bigger danger, which is why one of China’s most popular models is now banned from the U.S. government.

DeepSeek, a Chinese-owned and operated artificial intelligence company, develops several popular AI models, both in the frontier and in open-weight categories. DeepSeek V4 and V4 Pro are the company’s flagship options that are most comparable to GPT-5.5, Claude Opus 4.7, and Gemini 3.1, while DeepSeek-R1 is the open-source alternative that can run on local hardware.

DeepSeek doesn’t belong anywhere near the American people, much less federal civilian executive branch agencies.

It’s no secret that China wants U.S. companies to choose open-weight platforms like DeepSeek over American-made frontier models from OpenAI, Anthropic, and Google.

After all, they’re cheaper with no recurring subscription fees, data is isolated to local hardware instead of sharing it with the Big AI companies in the cloud, and with enough support, it could supplant U.S. dominance in the AI sector.

Some companies took the bait, too, with big names like Microsoft looking into ways to integrate DeepSeek into a subset of its products. So far, the Redmond giant hasn’t actually followed through, though, so no harm, no foul.

Corporations aside, DeepSeek also apparently had some kind of unofficial integration with U.S. government agencies. This was made apparent when the Department of Homeland Security issued a partial ban on the use of DeepSeek in federal civilian executive branch agencies. These include groups like the Department of Homeland Security, Department of Health and Human Services, Department of Energy, and many more.

As part of the ban, affected agencies must remove any trace of DeepSeek from their frontend and backend systems. Agencies must also remove any third-party software that utilizes or connects to DeepSeek in any way, including but not limited to products that use DeepSeek as part of their feature set or were developed by DeepSeek.

In the official order, DHS cited several clear reasons for the ban, all relating to national security:

• Data access: User prompts and interactions with DeepSeek — including classified or sensitive data — can potentially be sent back to servers in the People’s Republic of China, where it can be accessed by the Chinese government. This would potentially reveal U.S. government secrets and other intelligence to China.
• Cybersecurity: The DHS noted that DeepSeek’s “models themselves may contain malware or other security vulnerabilities.” Just as AI has been an effective ally in uncovering software security holes that were previously unknown, hackers have also used AI to spread viruses and commit crimes. So far, the DeepSeek app for iOS contains several critical flaws dating back to 2025, making it a security risk.
• Availability: Since DeepSeek hosts both frontier and open-source models, at least some version of the company’s AI is freely accessible, easy to implement, and virtually unregulated, presenting potential threats to the U.S. government.

Following the DHS’ lead, the Department of Defense and the director of National Intelligence issued their own DeepSeek bans to kick the AI out of U.S. military agencies, ensuring that the platform can’t gobble up important information about our national security and foreign operations.

RELATED: How to protect yourself and your data during the PHP web meltdown

amgun/Getty Images

It’s important to note that the ban only affects government agencies. DeepSeek is not blocked from public use by civilians or enterprise integrations via private businesses. These groups can still access DeepSeek without restrictions and, by affiliation, are still subject to DeepSeek’s influence.

That said, considering the platform’s deep ties to the People’s Republic of China, you may want to think twice before you install DeepSeek or use it on any of your devices.

So far, DeepSeek’s AI has been caught censoring itself regarding responses that paint China in a negative light. Back when the model made headlines in 2025, users noted that it would begin to respond to problematic questions involving controversial figures like Genghis Khan and Mao Zedong, only for the answer to disappear in real time.

In September 2026, DeepSeek was caught routing queries through Anthropic’s Claude in a large-scale distillation plot. Distillation is when an AI company uses query outputs provided by a larger, better AI model to train their smaller model to produce similar results for less. If imitation is the sincerest form of flattery, then DeepSeek must love Claude for copying off of its models.

Lastly, DeepSeek has the power to serve as a propaganda machine for the People’s Republic of China, swaying opinions on everything from U.S. politics to the overall perception of China and more. It doesn’t belong anywhere near the American people, much less federal civilian executive branch agencies. For now, though, the ban only covers the latter.

Like Blaze News? Bypass the censors, sign up for our newsletters, and get stories like this direct to your inbox. Sign up here!