Is DeepSeek Safe to Use? The Honest Answer
DeepSeek is free, genuinely capable, and one of the most downloaded AI apps in the world — it's also blocked on government devices in Italy, Australia, South Korea, and a large part of the US federal government. Both things are true at once, which is exactly why the question keeps coming up. Here's what's documented, what's alleged, and the calm, practical answer.
The core question with DeepSeek isn't whether the model is good — it's where your conversations end up once you send them.
A Genuinely Capable, Genuinely Cheap Chinese AI Model
DeepSeek is a Chinese AI company that burst into global attention in January 2025 when its R1 model matched much of ChatGPT's capability at a fraction of the reported development cost — roughly $5.5 million for its V3 model, compared to well over $100 million for comparable Western models. The app briefly overtook ChatGPT on both the Apple App Store and Google Play. Used through its hosted app or website, DeepSeek works the same way ChatGPT, Claude, or Gemini do: you type questions, get answers, generate code, draft text.
The concern that followed wasn't about model quality — it was about data. DeepSeek's own privacy policy discloses that it collects chat history, prompts, account details, device information, and even keystroke patterns, and stores this data on servers located in the People's Republic of China. Under China's 2017 National Intelligence Law, organizations operating there can be legally compelled to cooperate with state intelligence requests, without a requirement to notify affected users. That's the single fact driving most of the restrictions that followed — not a claim about the AI itself being unsafe to talk to.
Separating the Real Incidents From the Noise
The January 2025 database exposure was real, documented, and patched within about 30 minutes of disclosure.
Italy's data protection authority blocked DeepSeek from its app stores within 72 hours of opening its investigation.
Coverage of DeepSeek splits into security-vendor jargon on one side and "it's fine, relax" threads on the other. The documented record sits in between:
The Database Leak — Real
In January 2025, security firm Wiz found an exposed DeepSeek database containing over a million log lines, including plain-text chat history and secret keys, sitting open on the internet. DeepSeek secured it roughly 30 minutes after being notified.
The App Flaws — Real
Mobile security researchers at NowSecure found the iOS app disabled a standard Apple security protection and used outdated encryption. Both issues have since been largely patched.
Jailbreak Resistance — Real
Security testing by Cisco found DeepSeek R1 failed to block any tested jailbreak attempts; a separate Qualys evaluation found it failed over half of its tests, worse than most major competitors.
Government Restrictions — Real
Italy ordered a nationwide app-store block within 72 hours. Australia, South Korea, and a large share of the US federal government, including the Navy, NASA, and both houses of Congress, have restricted it on official devices.
Italy's data protection authority, the Garante, found that DeepSeek's own privacy policy plainly discloses that data may be stored outside a user's home country, in violation of GDPR safeguards. That finding led directly to DeepSeek's removal from the Italian App Store and Google Play — though users who had already installed the app before the ban have reported it continues to function.
The Calm, Practical Answer
The honest bottom line, reflected consistently across independent security researchers: DeepSeek's models are genuinely strong, but the hosted app is a real privacy risk for anything you wouldn't want a foreign government to potentially access. The risk isn't that the app will infect your device — the documented incidents are data-handling and jurisdiction issues, not malware. For casual, throwaway questions on a personal device, the practical risk is low. For anything involving personal, financial, health, legal, or work-related data, security researchers and multiple governments agree: keep it out of the hosted app.
- General knowledge questions and casual research
- Non-sensitive coding help and learning
- Anything you'd be comfortable posting publicly
- Work documents, client data, or anything under an NDA
- Financial, medical, or legal information
- Anything covered by GDPR or similar regulated-data rules
Frequently Asked Questions
Is DeepSeek safe to use?
For casual, non-sensitive questions on a personal device, the practical risk is low. For confidential, work-related, financial, or regulated data, most security researchers and several governments say no, because DeepSeek's hosted app stores data on servers in China, where national intelligence law can compel access to it.
Why have governments banned DeepSeek?
Primarily over data jurisdiction: DeepSeek stores user data in China, and Chinese law can require companies to cooperate with state intelligence requests. Italy, Australia, South Korea, and multiple US government bodies have restricted or banned it on official devices for this reason.
Is there a safer way to use DeepSeek's technology?
Yes. DeepSeek's underlying models are open-weight, meaning they can be run locally on your own hardware or through a Western cloud provider. Doing this avoids sending your prompts to DeepSeek's servers at all, which removes the data-jurisdiction concern, though it requires more technical setup.
More AI Privacy Guides, Written Clearly
Browse our full library of AI privacy and safety explainers, updated as the landscape evolves.
More from the creator
Explore two other projects from the same creator.