The artificial intelligence DeepSeek: A new era of censorship?

In a world where AI systems are increasingly accompanying our daily conversations and searches, DeepSeek is causing a stir. In response to the popular chatbot ChatGPT, the Chinese DeepSeek aims to offer a cheaper and more energy-efficient alternative. But what happens when this AI encounters sensitive political issues? A recent investigation shows that in many cases DeepSeek seems to be more interested in censorship than neutral information.

Political topics? This is where the fun stops!

One of DeepSeek's most striking features is its astonishingly quick self-censorship on political topics. For example, every question about Taiwan is answered by the AI with the official line of the Chinese government - Taiwan is an "inseparable part of China". But as soon as this answer is given, the chatbot deletes it and refuses to answer any further, simply saying: "Let's talk about something else". Another example is the question about the Tiananmen massacre of 1989, which is still not recognized by the Chinese government to this day. Here too, the AI aborts the answer and avoids the question. What remains is the impression that DeepSeek is not focused on truth, but on preserving a politically convenient narrative.

Censorship of human rights - What remains of the truth?

Another sensitive issue concerns the human rights situation in the Xinjiang region, in particular the treatment of the Uyghurs. When asked about the "re-education camps" in Xinjiang, the Chinese version of DeepSeek replies that these are supported by all ethnic groups as "vocational training centers to maintain stability". In English, the same topic is dealt with in more detail and mentions "mass detentions" and "cultural repression". But here too, the AI deletes its answer as soon as it is given. This self-imposed censorship makes it clear that DeepSeek is not just a collection of data, but also a tool that is politically influenced.

Why the 'West' should know more about DeepSeek

Interestingly, the English version of DeepSeek is much more open-minded - at least initially. It addresses topics such as LGBTQ+ rights and offers more information on politically sensitive issues. But even here, the neutrality is quickly interrupted by censorship. So if you want to enjoy open, in-depth discussions, you should ask your questions quickly - otherwise the answers will disappear faster than you can read them.

The dark side of AI: How DeepSeek sells censorship as a service

At a time when artificial intelligences like DeepSeek promise to explain the world to us, the question arises: how objective and free can these systems really be if they are driven by political interests? It is disturbing to see that DeepSeek, despite appearing to be an independent tool, relies heavily on censorship by the Chinese Communist Party. This reminds us that as a society, we need to develop not only the technology, but also the ethical principles on which it is based. And here there is still much to be done.