A DeepSeek Ban is Absolutely Coming

In partnership with

Learn AI in 5 minutes a day

This is the easiest way for a busy person wanting to learn AI in as little time as possible:

1. Sign up for The Rundown AI newsletter

2. They send you 5-minute email updates on the latest AI news and how to use it

3. You learn how to become 2x more productive by leveraging AI

Sign up to start learning.

The existence of DeepSeek AI makes absolutely no sense.

The entire United States government, and most recently the Supreme Court, have worked themselves into a tizzy over concerns about TikTok.

The issue is the app’s Chinese ownership. TikTok’s parent company, ByteDance, started in China, and American politicians have raised concerns about national security, the sharing of American data, and TikTok’s ability to shape narratives for years.

Against that backdrop, DeepSeek seems like a strange anomaly. The Chinese AI system surged onto the scene last month, shaving almost $1 trillion off the U.S. stock market in the process.

DeepSeek is essentially a clone of OpenAI’s most advanced chatbot. Although its creators haven’t disclosed exactly how they built it, a mounting body of evidence suggests that they used a process called "distillation" to essentially copy GPT-4 on the cheap.

DeepSeek quickly became one of the most downloaded apps on the Apple App Store, with tens of millions of American users.

That seems strange because, when it comes to privacy and security, DeepSeek is like TikTok on steroids.

First, the app’s data handling is a dumpster fire.

DeepSeek states clearly in its terms of service that it takes any data submitted to it and transmits it to servers in mainland China. That potentially includes any prompt or data that users enter into the app, as well as anything they give it permission to access, such as their photos or files.

ByteDance has gone to great lengths to try (unsuccessfully) to show that it’s keeping its data in America. It has partnered with companies like Oracle to store all of its American data stateside, for example.

It’s strange, then, to see DeepSeek waltz in, openly state that it’s taking reams of data and moving them to China, and face almost no resistance.

From a national security perspective, the app isn’t much better. As the writer Jim "The AI Whisperer," has shown, DeepSeek is all too happy to shape narratives about the Chinese Communist Party and its relationship with America.

Ask the app about things like the Tiananmen Square massacre, and it will feign ignorance. It does an excellent job of toeing China’s party line.

Why, then, was TikTok forced into what will likely result in a sale or merger with an American partner, while DeepSeek is allowed to do as it pleases?

The answer may lie in the speed with which new AI technologies roll out.

OpenAI famously demonstrated this with the launch of ChatGPT—an app that was intended to be little more than a user-friendly wrapper around the GPT-3.5 model but turned into a world-changing sensation that made average people acutely aware of AI.

Within months, ChatGPT had racked up hundreds of millions of users. Those users have shown that they’re all too happy to jump over to a new, free app like DeepSeek if given the opportunity.

Back in the day, it would take months or years for a new social network to attract users. That gave regulators time to figure out how they wanted to handle new entrants to the space.

In the world of AI, that timeline is greatly accelerated. Years become weeks, and months become days. Regulators and politicians may simply have trouble keeping up.

Mark my words, though—the DeepSeek ban is coming. Bipartisan legislation has already been introduced that would ban DeepSeek and other Chinese AI platforms from targeting American users. South Korea has banned it this week. More bans will come.

This kind of legislation will solve the problem of companies offloading American data to the Chinese government.

But it won’t fix the more existential problem of increased competition for American AI giants like OpenAI.

Although the DeepSeek app is hosted by its Chinese creator, DeepSeek has also released its own distilled versions of its model essentially for free.

Although the exact specifics around its training data remain unclear (probably because that data was stolen from OpenAI), anyone can take the distilled versions of DeepSeek and run them in a variety of places—even on a cell phone or other low-powered hardware.

Already, American companies like Perplexity have enthusiastically integrated these cheap models into their own offerings.

That means that even if the DeepSeek app is banned, the model will live on.

Chinese companies’ ability to use AI to shape American narratives or grab American data may be short-lived. However, the impact of cheap competition—and the need for American AI companies to innovate in order to beat it—will remain.