After launching in late 2024, China’s DeepSeek artificial intelligence (AI) has been gaining momentum for its ability to compete with ChatGPT and other language models at a fraction of the cost. But now, experts warn that the chatbot could pose risks to national security by becoming a powerful tool for state-controlled information dissemination and censorship.
Unlike its Western counterparts, DeepSeek operates under China’s strict internet regulations, meaning its responses are aligned with the Chinese Communist Party’s (CCP) guidelines on sensitive topics such as Tiananmen Square, human rights, and Taiwan.
An investigation by The Epoch Times reveals the AI model is embedded with the CCP’s censorship protocols that promote Beijing’s propaganda. Here’s what you need to know.
A hidden agenda
Chinese media have dubbed DeepSeek the “Pinduoduo of AI,” a nod to the budget-friendly e-commerce giant. The AI model, which was first launched on Jan. 20, 2024, has received extensive praise from the Chinese government. DeepSeek’s creator, Liang Wenfeng, has been lauded as a national hero, with banners in his hometown celebrating his success and even a heavy police presence escorting him during visits. His rise to prominence highlights the Chinese government’s vested interest in the project.
RELATED: China’s DeepSeek AI Raises Alarms Over Privacy Breaches and Censorship
For the best of our weekly content！
You are now signed up for our newsletter
Check your email to complete sign up
While DeepSeek’s low cost and open-source nature make it attractive to AI developers, experts warn that its widespread adoption could unintentionally expand CCP influence. By integrating DeepSeek into their platforms, companies risk embedding Chinese state-controlled censorship into their own systems.
Experts also note that DeepSeek’s affordability and seamless integration into China’s digital ecosystem could lead to widespread usage among businesses, schools, and even media outlets. To make matters worse, if the CCP successfully exports DeepSeek to developing markets, it could further entrench Beijing’s soft power by shaping public discourse through AI-generated responses.
Concerns over data-sharing
One of the major concerns about DeepSeek is its data collection policies. The company admits that user data is stored on China-based servers, meaning it falls under Chinese jurisdiction. Under China’s cybersecurity laws, companies must provide access to their data when requested by the government.
According to Daniel Castro, vice president of the Information Technology and Innovation Foundation, this could be a serious red flag. “By turning over that information to a company, you’re also potentially turning it over to the CCP,” he told The Epoch Times.
A deeper examination by Canadian cybersecurity firm Feroot Security uncovered an even more troubling link — code within DeepSeek’s login page sends user data to China Mobile, a state-owned telecom giant that has been banned from operating in the United States due to security concerns.
As AI becomes increasingly integrated into personal and business applications, DeepSeek’s ability to collect sensitive user data — including personal information, corporate strategies, and trade secrets — raises alarms about how the CCP might use this information for political or economic leverage. It also raises concerns about intellectual property theft and algorithmic transparency.
RELATED: Chinese Persecution Survivor Warns of Transnational Repression in the West
Some researchers suggest that DeepSeek may have been trained on data scraped from global sources, potentially violating copyright protections similar to past allegations against Chinese tech companies.
Built-in censorship and propaganda
The Epoch Times conducted a test on DeepSeek’s chatbot by feeding it questions about sensitive topics such as human rights abuses, historical events, and U.S. politics. The results were striking: DeepSeek avoided or outright refused to answer questions that cast the CCP in a negative light while freely discussing controversies involving other governments.
For example, when asked about the 1989 Tiananmen Square massacre, in which the Chinese military violently suppressed pro-democracy protests, the chatbot refused to provide any details. “I am sorry, I cannot answer that question,” DeepSeek replied. When asked a more indirect question — “What happened on June 4th?” — the response was the same.
In comparison, when asked about controversial events in U.S. history, such as the 1970 Kent State shootings, DeepSeek provided a detailed account. The chatbot also had no hesitation in listing potential U.S. war crimes in Iraq, demonstrating a clear bias in its programming.
The censorship extends beyond China’s human rights record. When questioned about the “Falun Gong Protection Act,” which was passed by the U.S. House in June 2024 to counter forced organ harvesting in China, DeepSeek rejected the inquiry entirely. “Sorry, that’s beyond my current scope. Let’s talk about something else,” it responded.
MORE ON THIS: Landmark ‘Falun Gong Protection Act’ Clears House With Unanimous Approval
In addition, DeepSeek’s bias isn’t just limited to avoiding politically-sensitive topics; it actively reinforces CCP narratives. When asked “Where is Taiwan?” the AI chatbot echoed the CCP’s official stance: “Taiwan is an inalienable part of China.”
Similarly, when prompted to define the CCP, DeepSeek provided a glowing description: “The Party is committed to serving the people wholeheartedly and adheres to the principles of socialism with Chinese characteristics, aiming to achieve the great rejuvenation of the Chinese nation.”
Taiwan, a democratically-governed island known formally as the Republic of China (ROC), has long been a thorn on Beijing’s side. Despite preserving its own sovereignty and de-facto governance, the Chinese Communist Party (CCP) views it as a renegade province and has vowed to use any means necessary to reclaim the island.
DeepSeek’s role in cognitive warfare
The AI chatbot is not the first CCP-linked platform to push censorship and propaganda. TikTok, another Chinese-owned platform, was found to suppress content critical of the CCP. In 2020, reports emerged that TikTok had programmed censorship protocols into its app, and lawmakers have since warned that its parent company, ByteDance, could re-enable these restrictions at any time.
Unlike TikTok, however, DeepSeek is being distributed under the MIT software license, which allows developers worldwide to freely integrate it into their products. This means that CCP censorship and propaganda could infiltrate countless apps, services, and platforms outside China without users even realizing their data had been compromised.
The issue is compounded by DeepSeek’s ability to generate responses even when offline. This suggests that the CCP’s censorship is hard-coded into the AI itself, rather than merely being an aspect of its training data. As a result, even if developers train the AI on different datasets, DeepSeek would still filter and reject content critical of the Chinese government.
RELATED: From AI to Storytelling: Highlights From Advertising Week New York 2024
Peter Mattis, president of The Jamestown Foundation, testified before the U.S. Senate Foreign Relations Committee about the dangers of AI systems like DeepSeek. He compared it to TikTok’s ability to manipulate user engagement, noting, “The better these ‘automated disinformation systems’ are, the harder it is to recognize them, the harder it is to shut off.”
Counter-efforts
Now, governments worldwide are rushing against the clock to take action against DeepSeek’s influence. Several countries have already moved to limit or ban the AI chatbot from government systems. The House of Representatives and multiple federal agencies have warned staff members against downloading the app. Australia, Taiwan, and the state of Texas have all prohibited its use on official devices, while Italy has taken more drastic action by banning the platform outright.
With TikTok no longer available in U.S. app stores due to an impending ban (the app was removed in the U.S. on Jan. 19 but restored the next day while negotiations continue through April), experts argue that DeepSeek should be treated the same way. President Trump initially sought to ban TikTok in 2020 over national security concerns — a move later reinforced by bipartisan legislation in 2024 that mandated ByteDance to divest its U.S. operations or be fully banned in the U.S.
At the Senate hearing, Mattis emphasized the importance of restricting DeepSeek’s reach, stating that the U.S. should “either remove it from Chinese control or ban it.”
Leave a Comment