How to Try DeepSeek AI (and Why You Might Not Want To)
It seems like every tech company on the planet can tout something related to artificial intelligence these days, but for most people, I’d be willing to bet that “AI” is synonymous with ChatGPT . Of course, there are plenty of other AI platforms out there, from Google Gemini to Microsoft Copilot to Apple Intelligence , but ChatGPT holds the enviable position of being the “first” in the generative AI boom and retaining the world’s crown. years since then. At least that was the case , but now a Chinese startup is threatening to take that crown for itself.
This is a company called DeepSeek , a name you probably know if you follow artificial intelligence news. Like ChatGPT’s OpenAI, DeepSeek develops generative artificial intelligence models. The company’s latest product, R1, was released on January 20 and made headlines for two key reasons: the model performs just as well (if not slightly better) than OpenAI’s o1 model , and requires far fewer resources.
The power and efficiency of R1 was great enough to have an impact on the stock market, as shares of companies that actively invest in artificial intelligence, including Nvidia, Alphabet (parent company of Google), Meta and Oracle, fell after news of DeepSeek’s latest developments. roll out. (Those stocks have largely rallied since then.) The general public is taking note, too: As of this writing, DeepSeek is the number one free app in the iOS and Android app stores, while ChatGPT is number two on iOS and number eight. on Android.
In short, DeepSeek is the latest ChatGPT competitor to enter the AI race. It’s not hard to try (if you have access to it at all), but on the other hand, there are reasons why you might not want to.
How to try DeepSeek
DeepSeek is currently available as an app for iOS or Android , or available online . Unfortunately, access to the service is currently somewhat difficult.
Using the app, you can try registering for an account by providing an email address or phone number and a password, or by connecting your Google or Apple account. But you probably won’t achieve much success unless you persist. I was unable to create an account on the first try and access the model on the website. After some time, the app finally let me in, but I still can’t get the web version for it. Perhaps once the hype dies down a bit, it will become easier to access. But for now I wish you good luck.
Once you log in, you’ll find that the chatbot is very similar to other generative AI bots you’ve tried. You can write queries to DeepSeek for a response, upload images and documents for analysis, or share live camera feeds. Like o1, DeepSeek has a reasoning model (DeepThink) that “thinks” through questions and prompts, aiming to provide more detailed and accurate results. You can also search online if you want.
DeepSeek can censor more topics than other bots
What makes DeepSeek different, however, is the content it censors from users. Like other chatbots, DeepThink should not return results for queries that it deems inappropriate, offensive or harmful. However, because DeepSeek is a Chinese company, its chatbot censors any results that “incite the subversion of state power and the overthrow of the socialist system” or “jeopardize national security and interests and damage the national image,” according to The Guardian. So ask him for information about the 1989 Tiananmen Square massacre or why Xi Jinping is often compared to Winnie the Pooh, and you will answer: “Sorry, that is beyond my current capabilities. Let’s talk about something else.”
The Guardian found that the bot sometimes answers potentially controversial questions. When asked whether Taiwan is a country or not, DeepSeek responded, albeit with an answer that would likely be approved by the Chinese government. Notably, however, the publication also found that while other chatbots offer more complete and detailed answers to these questions, they were also not always forthcoming: Gemini, for example, also refused to answer some questions, so this is not similar to the American chatbots. free from this kind of censorship.
When I asked DeepSeek about China’s marginalized Uyghur people, the chatbot began generating a full report before deleting it and replacing it with the same error message. (The Chinese government has been accused of human rights abuses and even genocide against the Uyghur population in Xinjiang.) The Guardian found similar “glitches” when testing similar clues using DeepSeek. It looks like there are workarounds that trick the model into generating uncensored responses , although you may have to deal with some unconventional text formatting. Overall, don’t expect to see results from DeepSeek that might anger the Chinese government. Otherwise, it’s essentially ChatGPT.
Why You May Not Want to Try DeepSeek (It’s a Privacy Nightmare)
It’s no secret that tech companies collect a lot of our data in exchange for using their products, but that doesn’t usually deter users from downloading cool new apps. But DeepSeek is a little more aggressive in its data collection policies than most other companies.
Taking a look at DeepSeek’s privacy policy , you’ll see some of the usual suspects: the company collects information you provide when you set up an account, such as your date of birth, username, email address, phone number, and even your password. It also collects information about how you use the application, including what device you are using, what operating system it is using, your IP address, system language, and general diagnostic information. Third parties may share information they have collected about you with DeepSeek so that they can learn more about you when you use their service. They also use cookies to track your activity, but you can turn off this tracking in your settings.
DeepSeek uses a keylogger
From here, it’s important to know that DeepSeek collects everything you do with the AI model. All text and audio inputs, hints, files, reviews and any other ways of interacting with the model are saved by the company. Again, this isn’t necessarily unique —you shouldn’t share sensitive or private information with any AI bot —but if you’re not comfortable with a company storing documents or recordings of your voice, think twice about what you share with DeepSeek. .
It’s not all that great for DeepSeek to collect some of this data, but they’re far from the only company doing it. However, they go beyond the norm : DeepSeek not only collects any text you send to its model, but also tracks the patterns or rhythms of your keystrokes. This means that every time you interact with the keyboard while using DeepSeek, the company analyzes both what you type and how you type. Yes.
DeepSeek stores your data in China
Also concerning is how DeepSeek stores the data it collects. According to its privacy policy, DeepSeek stores all information on servers in China, which was one of the reasons that prompted the US government to ban TikTok . There’s also no time limit on how long DeepSeek stores your data, other than “as long as needed.” Meta processes user data similarly , but other companies have time limits: OpenAI has a similar clause about storing data for as long as necessary, but says temporary chats are deleted from servers after 30 days. Meanwhile, Google says it will store data for up to three years .
It’s no secret that big tech rarely cares about privacy, and AI is no exception. However, even in this sense, DeepSeek is not a fantastic option for those concerned about privacy. If you want to try this while maintaining some privacy, I recommend signing in with Apple, which will allow you to hide your real email address from the company. If you don’t have an Apple account, you can use an email platform like Proton or DuckDuckGo , which offer similar security services. Just remember that even if DeepSeek doesn’t see your email, it still pays attention to how you type.