Are you using the internet? Of course you are because you have one of the best computers for using it right in your hands. Since you are using it, you have no doubt seen people talking about DeepSeek AI, the new ChatBot from China that was developed at a fraction of the costs of others like it.
Like most things you read about on the web, this is not something you should dive into blindly. There are good reasons to want to try it; there are also good reasons not to. We're going to talk about all of them.
What is DeepSeek AI?
DeepSeek is a Chinese AI laboratory that has recently released the DeepSeek R1 AI model. It's an AI ChatBot based on Meta's free and open-source Llama 3.3, trained by the DeepSeek team.
You're going to read a bunch of terms like LLM (Large Language Model) and reasoning, but what it all means is that researchers and engineers worked on writing software that can be "trained," either through manual input or by actually searching the internet, to find the answer to a question and present it in a way that sounds like a real person wrote it. The software can separate different problems from a single question and tackle them individually. It's pretty cool.
You've probably used something like ChatGPT or Google's Gemini Live. DeepSeek R1 is just like that, and some people think it's even more accurate than OpenAI's latest ChatGPT reasoning models.
DeepSeek is good, but it's not perfect. It will return incorrect results and it's censored by the Chinese government. If you do try it, don't bother asking about the Uyghur people or Tiananmen Square.
Why is everyone talking about DeepSeek AI?
There are two reasons: it's new, and it costs a lot less to develop than similar AI models. A lot less.
As an end-user who is not an investor or Nvidia shareholder, I don't care about the costs. While it was far less than the amount OpenAI spent, it's still an astronomical amount that you or I can only dream of having access to. Technology is that way — companies choose to spend money to develop something instead of feeding starving kids. It's not going to change.
I will say one thing about the cost: Is anyone surprised that China was able to produce something of comparable or greater quality than a Western product at a fraction of the cost? Labor costs, taxes, environmental protections, and bureaucracy all add up. China has all of this figured out for the time being.
Other than the price, the simple fact is that DeepSeek R1 is new and works well. It debuted in December 2024, quickly gained popularity in mobile app stores, and now everyone is talking about it. It's the very same thing that happened when ChatGPT was first made public. We find new novelties interesting as a species.
What about privacy issues with DeepSeek AI? Should you try it?
Yes, there are plenty of privacy issues, and the same issues you would find with the same type of products from Google, Apple, Microsoft, etc., apply to DeepSeek. You are entering data into the machine every time you type in the box. How that data gets treated is very importan,t and companies need to be fully transparent about how it is collected, used, and stored.
DeepSeek is actually pretty good with this. The company's privacy policy spells out all the terrible practices it uses, such as sharing your user data with Baidu search and shipping everything off to be stored in servers controlled by the Chinese government.
You can download and install DeepSeek locally to bypass some of this data collection, but not all of it. In any case, the company is likely betting that you either won't care or just won't read the privacy policy. I applaud both their audacity and transparency.
There is speculation that DeepSeek is also doing some very sketchy things, especially through the mobile app. Google, Microsoft, OpenAI, and META also do some very sketchy things through their mobile apps when it comes to privacy, but they don't ship it all off to China. Our official company guidance from IT is not to install it on a work-provided device and to not use our company Google accounts to sign into DeepSeek's website. The U.S. Navy has stated that the service can not be used in any capacity by any service member.
Better safe than sorry. You need to think about these issues before you try DeepSeek. Don't let the hype and fear of missing out compel you to just tap and opt-in to everything so you can be part of something new. Not because DeepSeek comes from China, but because you should do this for every new awesome thing you read about on the internet.
Is this going to get banned in the United States?
It's possible, and it's already being investigated.
White House press secretary Karoline Leavitt says that DeepSeek is being investigated to determine the national security implications of the service, and the administration's cryptocurrency czar, David Sacks, claims intellectual property theft could be happening. OpenAI is saying the same thing, claiming that DeepSeek secretly used its software.
President Donald Trump says this should be a "wake-up call" to the American AI industry and that the White House is working to ensure American dominance remains in effect concerning AI.
DeepSeek sends far more data from Americans to China than TikTok does, and it freely admits to this. The data is also potentially more sensitive as well. It's hard to say if someone in Washington will decide that DeepSeek is abusing our data or causing U.S. companies to lose too much money and ban the service outright. It might happen, and we shouldn't be surprised if it does.
