A new AI product from China, called DeepSeek, has been making waves. With DeepSeek R1 as its chatbot, it’s certainly an unfortunate name, but that’s a cultural thing. DeepSeek is getting a lot of attention because AI and because China. Additionally, it’s claimed that DeepSeek has much lower development costs compared to mostly American competitors like OpenAI (the creators of ChatGPT). Whether this is truly the case remains to be seen, quite likely the Chinese government lent a helping hand. Anyway, three reasons not use DeepSeek;
Data Risks
The Chinese government pursues a different societal order than what we are accustomed to in the Western hemisphere. That’s their prerogative, but it has certain consequences that we’re not entirely comfortable with. For example, when you install the Temu app, you most likely agree to the terms without much thought. If there was something wrong with it, it would be banned, right? Wrong. Other than the privacy violations we’ve gotten used to from all our apps such as allowing access to your camera and microphone and viewing your contacts and photos, you also agree that Temu can activate your phone (at night), read your messages, and take screenshots of them. When reporters asked questions about this Temu’s response was; “Well, we don’t do that”. Which is obviously not convincing at all. If you don’t plan on using it, why ask for permission? At the very least, this clearly shows that if they want to, they can take over your phone (which, let’s face it, contains your enire life nowadays) from the other side of the world. If Temu ever decides to stop selling stuff, they have all the information they need to take everything from you—your money, your house, your identity. They could even do this on a large scale with a bit of help from AI, like DeepSeek’s AI.
DeepSeek emphasizes that their model is fully open-source and free to download on your own device, theoretically meaning your data isn’t shared with them. But asTemy has demonstrated, the data you share while interacting with the app might not be the most important thing you have to worry about.
App builders can take control over your smartphone, which effectively means your life, from the other side of the world.
Censorship
There are no completely objective AI chatbots; they all have some bias in a specific direction. Usually, this bias is quite subtle, but not in the case of DeepSeek. The DeepSeek chatbot is clearly censored on topics the Chinese government prefers not to be open about. Things like ‘Taiwan’, the events at Tiananmen Square in 1989, or the protests in Hong Kong. It’s a bit clumsy to apply such crude and visible censorship on topics users don’t need a chatbot for and journalists will definitely look into. If such obvious censorship is applied, it’s very likely that the DeepSeek bot will also be biased favorably for the Chinese government in other ways.
This will not have an enormous short term impact. But as we are seeing with the rising of Extremist right wing populism in Western societies; the revolution will not happen with one big event, but with a continuous stream of small events.
Cheaper. Not better.
From various online reports, it seems that DeepSeek performs better in calculations than OpenAI and that OpenAI beats DeepSeek in creative tasks and news generation.
However, DeepSeek is significantly cheaper for users compared to OpenAI. DeepSeek charges about $0.14 for the same amount of text (around 750,000 words) for which OpenAI charges $7.50. Business users with a $20 per month OpenAI subscription could manage with a free DeepSeek subscription in terms of text volume but with less creativity. In the end, this still seems like a small price to pay.
However, I often see self-proclaimed keynote speakers (seriously, what sane person would refer themselves as a keynote speaker?) and business gurus on LinkedIn praising the likes of Temu and other Chinese business models. Completely ignoring (or being completely ignorant which is also plausible) the human rights and environmental violations, the consumer manipulation and poor product quality. From the many enthusiastic responses they receive, it is clear that there are many people shortsighted enough to give up everything of longterm value, to make an extra Euro today.
Mistral
Just likely mostly everyone, I have been comparing DeepSeek to OpenAI here. OpenAI’s ChatGPT made AI known to the general public and for many people it is synonymous with AI. But obviously, it isn’t, there are more options out there.
Among all these options, I wouldn’t choose OpenAI either. In recent weeks, we’ve clearly seen that European interests and ideals aren’t necessarily well-protected by the American governmental or Big Tech companies. even though Europe is far from perfect, it is closest to the democratic, fair and free world where I would want to live.
Fortunately, there’s a wonderful European alternative to OpenAI: the French Mistral. Their AI chatbot, Mistral 7B, is just as open-source and free as DeepSeek R1, just not as scary.
Erwin Fisser 