Washington and Europe are growing wary of DeepSeek.
The Chinese artificial intelligence company astonished the world last weekend by rivaling the hit chatbot ChatGPT, seemingly at a fraction of the cost. But now, regulators and privacy advocates are raising new questions about the safety of users' data.
Regulators in Italy have blocked the app from Apple and Google app stores there, as the government probes what data the company is collecting and how it is being stored.
In France and Ireland, officials are digging into whether the AI chatbot poses a privacy risk.
And in the U.S., members of Congress and their staff are being warned by the House's Chief Administrative Officer not to use the app. Two U.S. Representatives recently asked the Trump administration to strengthen existing restrictions on the sale of semiconductor chips to China in an effort to "outcompete" China in AI development and "safeguard Americans' data."
What data is DeepSeek collecting?
According to DeepSeek's privacy policy, the service collects a trove of user data, including chat and search query history, the device a user is on, keystroke patterns, IP addresses, internet connection and activity from other apps.
Other AI services, like OpenAI's ChatGPT, Anthropic's Claude, or Perplexity, harvest a similar volume of data from users. So do social media apps like Facebook, Instagram and X. At times, these kinds of data collection practices have led to questions from regulators.
"Data security concerns are always a critical issue when using AI chatbots, and this is not unique to DeepSeek," said Angela Zhang, a law professor at the University of Southern California who specializes in Chinese regulation. "Even U.S.-based AI firms like OpenAI have faced significant scrutiny and investigations in the EU over potential data privacy violations," she said, referring to the European Union's strict data protection laws. For example, these require users to opt in to any data collection.
Where does DeepSeek store user data?
DeepSeek sends all the data it collects on Americans to servers in China, according to the company's terms of service.
And for cybersecurity experts, that is where the problem lies.
Much like Washington's fears about TikTok, which prompted Congress to ban the app in the U.S., the concern is that a China-based company will ultimately be answerable to the government, potentially exposing Americans' sensitive data to an adversarial nation.
(The Trump administration is currently not enforcing the TikTok ban. The president signed an executive order seeking to extend the ban's start date until April.)
When it comes to DeepSeek, Samm Sacks, a research scholar who studies Chinese cybersecurity at Yale, said the chatbot could indeed present a national security risk for the U.S.
"That data, in aggregate, can be used to glean insights into a population, or user behaviors that could be used to create more effective phishing attacks, or other nefarious manipulation campaigns," Sacks said.
Has the Chinese government accessed Americans' data through DeepSeek?
There are no public reports of Chinese officials harnessing DeepSeek for personal information on U.S. citizens.
Much like with the debate about TikTok, the fears about China are hypothetical, with the mere possibility of Beijing abusing Americans' data enough to spark worry.
Yale's Sacks said there are two other major factors to consider about the potential data risk posed by DeepSeek.
First, the Chinese government already has an unfathomable amount of data on Americans.
In December, Chinese hackers breached the U.S. Treasury Department's computer systems. Last year, another group of Chinese hackers spied on Americans' texts and calls after infiltrating U.S. telecom companies. Not to mention that an enormous amount of data on Americans is routinely bought and sold by a vast web of digital data brokers.
"China is already highly capable of gaining access to Americans' data, even without a potential backdoor like DeepSeek," Sacks said.
And secondly, DeepSeek is open source, meaning the chatbot's software code can be viewed by anyone. Developers can also build their own apps and services on top of the underlying code.
Most popular AI chatbots are not open source because companies closely guard the software code as confidential intellectual property. Sacks argues that DeepSeek providing transparency into how data is being accessed and processed provides something of a check on the system.
Yet American privacy experts still don't know what happens to the data once it is stored on servers in China, where DeepSeek would be subject to national security laws that would force the company to hand over data if the government made a request.
Requests for comment to DeepSeek, and its owner, Chinese hedge fund High-Flyer, were not returned.
Does DeepSeek censor answers?
Researchers and reporters have found examples of DeepSeek both pushing Chinese propaganda and censoring sensitive political subjects.
While competing chatbots have no trouble explaining the 1989 Tiananmen Square massacre, DeepSeek told NPR: "Sorry, that's beyond my current scope. Let's talk about something else."
Likewise, when NPR asked about the sovereignty of Taiwan, which China claims as part of its territory, DeepSeek started writing an answer including that Taiwan is a "complex and contested issue" before the answer disappeared and was replaced with the message that the question is beyond DeepSeek's scope.
Experts say the chatbot's limits are a reminder that the internet is government-controlled in China and that tech companies like DeepSeek are subject to interference, so the chatbot's answers should be greeted with a heavy dose of skepticism.
Is there any safe way to use DeepSeek?
Privacy experts like Sacks say there are ways of using DeepSeek more safely.
They suggest not signing up to the app through Gmail or Apple accounts to ensure data on those services remains unavailable to DeepSeek.
And when using the chatbot, it's best to not ask questions revealing identifying or highly personal information, since all inputs into DeepSeek are used to train the AI model, data that is eventually stored in China.
People can also use a virtual private network, a VPN, to hide their location.
If you have a government job, or if your personal information is potentially sensitive, she said, avoiding DeepSeek altogether is the safest bet.
Ultimately, though, in the U.S., the task of safeguarding digital data is up to users, since there are no federal data privacy laws on the books, unlike the European Union, which has laws giving citizens insight and control over how their digital data is being accessed and used by private companies. Instead of creating national regulations, Sacks said, "We're kind of playing Whack-A-Mole with this rotating cast of apps from China."
She continued: "We can ban TikTok, we can raise the red flag of data security concerns for DeepSeek, but we don't have a more comprehensive way to ensure that companies are not collecting more data than they should."
Copyright 2025 NPR
