DeepSeek R1 is the hot new artificial intelligence chatbot on the scene. Developed in China, it’s a powerful ChatGPT competitor that’s taken the AI world by storm. It’s much more efficient than the US’ company offerings, with OpenAI, Meta, and Google now racing to outdo this free, open-source alternative.
However, in 2025, it’s always good to side with caution. The AI chatbot might be good at presenting reasoning for an answer and be entirely free, but what’s the big catch? Well, so far, it doesn’t seem to differ that much from the safety precautions you’d take with any other app.
DeepSeek R1 does request your data, which will presumably be stored on Chinese-based servers. This is no different from Google or OpenAI requesting your data if you log into them, except the data is stored in the US.
For additional precautions, you can use Hugging Face to use the AI in a sandbox environment. While it won’t be as fully featured as the downloadable app, it can alleviate some of the concerns of using an app like it. If you want to lock down all the way, you can even host it locally if you have the necessary hardware to run a large language model.
Obviously, as with any LLM web interface, don’t input sensitive information into the chat. DeepSeek could still be training off of its own responses and interactions with humans, and if that data is hoovered up to train it, it could lead to consequences down the line.
DeepSeek R1 appears safe so far outside of government censorship
DeepSeek R1, from our own poking around, appears to be as safe to use as any other app of its ilk. We are keeping tabs on security experts talking about it across social media, but nothing sticks out as of yet. The main difference you’ll find is that, as with any piece of software made in China, it’s heavily censored to appease the government.
This means you can’t ask it to infer something along the lines of insulting Xi Jinping or ask it about Tiananmen Square. However, Google Gemini and ChatGPT have struggled in the past to say whether Palestinians are free, which is now replaced with a long explanation in recent queries.
DeepSeek is also open source. If you’ve major concerns about the potential safety problems from software like R1, you can wait for more analysis to come out. However, so far, nothing to do with safety concerns has been brought up by the open-source community.