In spite of a turbulent period of emergence, popularity, cyberattacks, and outages, the DeepSeek AI platform has taken a firm grip on the technology world.
Experts have already pitted DeepSeek against ChatGPT to see if the new kid on the block holds its own against more experienced AI. But, the question remains, performance-aside, is the platform actually safe?
In the wake of the US TikTok ban, it would appear DeepSeek offers several concerning similarities to the social platform in the form of its privacy policy, concerning app activity, and the location of its servers. What's more, the service offers its capabilities at a much cheaper price, so if you're financially better off, what cost are you paying instead?
We've taken a closer look at the platform to help you decide whether you should risk your data for a new, cheaper, AI future, or if, in fact, the concerns aren't quite as alarming as initially thought.
A not-so-private policy
A good privacy policy would traditionally be a source of reassurance for the privacy-conscious. In DeepSeek's case, this isn't so. The privacy policy reads:
"When you use our Services, we may collect your text or audio input, prompt, uploaded files, feedback, chat history, or other content that you provide to our model and Services...the personal information we collect from you may be stored on a server located outside of the country where you live. We store the information we collect in secure servers located in the People's Republic of China"
The result of this is that, in reality, DeepSeek could be collecting:
- Your name, date of birth, and email address
- Data inputted into the platform. Whether that be audio, text, or previous chats.
- Technical information surrounding your device – including IP, operating system, and keystroke patterns
As the policy states, this information is then stored on servers in China. The transfer of this data to China then raises questions regarding what happens next. The privacy policy outlines the data can be used to "Review, improve, and develop the Service" as well as "Comply with our legal obligations, or as necessary to perform tasks in the public interest, or to protect the vital interests of our users and other people." and, of course, advertise.
What's more, the provider claims to retain information for "as long as necessary" and doesn't state any exact periods by which all of your information may be removed.
Given the nature of this information, and how it's used, there are legitimate concerns about the long-term risks to your data and the potential non-existence of true privacy.
Unexpected activity in the AI app
Concerns about the links to Chinese authorities don't stop there. One X user pointed out some rather unusual activity from the AI service when discussing topics seen as sensitive in the region.
There's no way of proving this means DeepSeek is in any form of continued relationship with authorities, though it does raise questions about the nature of information received on the platform. However, this is nothing new for AI. Who remembers the great glue on your pizza fiasco?
DeepSeek’s newest AI model is impressive—until it starts acting like the CCP’s PR officer. Watch as it censors itself on any mention of sensitive topics. 🧵👇 1/9 https://t.co/nN6EwHsnoeJanuary 20, 2025
Even more cause for concern
The news surrounding DeepSeek since its launch a little over a week ago has not helped its case. The service was hit by a large-scale cyberattack, which raised questions about its security and forced DeepSeek to limit registrations for the time being. The DDoS attack still appears to be impacting the service though the move to limit registrations comes to ensure core functions remain in operation.
What's more, Italy has launched an investigation into DeepSeek over privacy concerns. The country has launched the investigation due to the "Possible risk to the data of millions of people in Italy," according to the Garante, Italy's data watchdog authority.
In amongst the news flurry surrounding DeepSeek, it's gone rather unnoticed that cybersecurity firm KELA was able to jailbreak the service. In a new report, KELA outlines several key vulnerabilities found within DeepSeek that "revealed that the model can be easily jailbroken using a variety of techniques, including methods that were publicly disclosed over two years ago."
What do the experts say?
Given the rapid growth of DeepSeek in the AI space, experts are understandably concerned. Speaking to TechRadar, Cybersecurity expert at NordVPN, Adrianus Warmenhoven, commented "that DeepSeek's AI model has faced growing backlash for its refusal to address political topics. This has sparked concerns about potential biases and external influence on the platform's content moderation policies."
On the concerns surrounding its privacy policy Digital Rights Expert Lauren Hendry-Parsons said "There are genuine concerns around the technological potential of DeepSeek, specifically around the terms of its privacy policy. It states that users’ keystroke data will be collected, along with their IP address, and this will be combined with information from other sources such as mobile identifiers, cookie identifiers, and more."
Want to know more? Read “This is a wake-up call" - the DeepSeek disruption: 10 experts weigh in
Why risk it at all?
While nothing has explicitly gone wrong as yet, it's advisable to approach DeepSeek with caution. In online privacy, if something seems oddly cheap, you're usually paying with something other than money - like your sensitive personal data.
With more trustworthy alternatives such as ChatGPT and OpenAI available, and very capable, it's a simple choice to pay more to be secure. While Trump is yet to show concern about the privacy issues surrounding DeepSeek, it wouldn't be surprising if such a platform were to meet similar opposition to TikTok.
