Cheap and efficient are two pretty compelling attributes and are drawing intrigued AI enthusiasts from around the world to try out open source generative AI model DeepSeek.
The Chinese-owned venture is enjoying a massive boom in customers, even, ironically, as the US Government warns about potential national security issues that could come with using a Chinese digital service.
But, AI LLMs need data to train on, so what customer data is DeepSeek collecting and sending to China? And what shouldn’t you share with the AI platform?
What Data Is DeepSeek Collecting?
DeepSeek does have an English guide online as to what data it collects. It’s quite a list. It includes your date of birth; email address, telephone number and the password you set up. This is pretty standard. Also, this is a great opportunity to remind you that your password should be different across every platform you use!
However, as it’s an AI model, you will also be handing over your “text or audio input, prompt, uploaded files, feedback, chat history, or other content that you provide to our model and Services”.
This just in! View
the top business tech deals for 2025 👨💻
The terms also add that if you, as a customer, contact the Chinese-venture, they will ask for “proof of identity or age”, and this data they will keep.
More Data that DeepSeek Collects
Your IP address, unique device identifiers, and cookies will also automatically be shared when you use DeepSeek. The identifiers include “your device model, operating system, keystroke patterns or rhythms, IP address, and system language”. Crash reports and performance logs are also collected.
If you pay for any services from DeepSeek, your payment details will also be saved and this is for “customer services” and “after-sales support”. So far, again, pretty standard.
The last category is “information from other sources”. For example, if you log onto DeepSeek using an Apple or Google sign-in, the company will receive some information from them. And there is also a note about advertisers who share information with DeepSeek, including “the products or services you purchased, online or in person” outside of the service.
Wired has looked into this “underlying activity” and says that DeepSeek is sharing data with the “Chinese tech giant Baidu, potentially for web analytics purposes, as well as Volces, a Chinese internet infrastructure firm”.
Should You Be Concerned About DeepSeek?
The big difference between OpenAI and DeepSeek is not the data that the companies are collecting – in fact they are doing similar things – but where this data ends up. DeepSeek is sending the data it collects to China and is completely open about this. It states: “We store the information we collect in secure servers located in the People’s Republic of China.”
It adds the pretty standard legalize terminology that it will share data to “comply with [its] legal obligations” but in this case, it means that data can be shared at any time with Chinese government officials. This means DeepSeek’s data can be used in accordance with a 2017 law, stating that organizations must “cooperate with national intelligence efforts.”
The Chinese government has the right to access data collected from US DeepSeek users and can use it for allegedly nefarious purposes. These could include pushing Chinese propaganda, which is an accusation that has been levelled at TikTok. In fact, some DeepSeek users are already reporting that some answers the AI returns sound like propaganda and some questions relating to topics deemed sensitive by the Chinese government – for example the Tiananmen Square massacre – are returning blanks.
What Not to Share on DeepSeek
The big question is whether you think the fact that all of this data going to China is a problem for you. You have probably heard it all before, but there are already some concerning warnings about how this date may be manipulated for the benefit of the Chinese benefit of the Chinese regime. These include some dire predictions about cyber threats that directly target the West, utilize social engineering, and exploit vast quantities of sensitive data.
The fact that TikTok had around 170 million active monthly US users suggests that many people are not too concerned about these threats, despite repeated warnings from the government. This has absolutely been borne up by the explosion in use of another Chinese social media platform, RedNote, as the TikTok ban loomed.
If this isn’t a concern for you – and you are also aware of the data that AI models will be capturing as standard – then just take some sensible precautions.
- If you are a creator, as we warned with ChatGPT, if you share your original works, be prepared for them to be shared with another user
- Don’t input any sensitive or personal information into the AI assistant, such health or personal data
- Don’t use it to store your usernames/passwords
- Be aware that you can also delete your chat history in DeepSeek through the Settings option
- Don’t use the AI models if you are using a work network
- Don’t upload company sensitive documents or client information
It’s also worth noting that installing the app on your computer allows a greater level of privacy control, but if you’re unsure about it in the first place, having the app sit on your PC is unlikely to appeal.
Ultimately, if TikTok’s fate is anything to go by, the US government is quite likely to take action against US data flowing freely to China. TikTok’s data servers for US customers were not, after all, in China but it was still deemed enough of a threat to be banned.
While new AI offerings are fun – especially at low cost and with impressive capabilities – US users might soon be forced to return to “safer” options – even with their data grabbing tendencies. And this is not least because OpenAI and Anthropic have serious US money behind them and DeepSeek does not.
