2 Sources
[1]
DeepSeek's answers include Chinese propaganda, researchers say
Since the tool made its debut this month, rattling stock markets and more established tech giants like Nvidia, researchers testing its capabilities have found that the answers it gives not only spread Chinese propaganda but also parrot disinformation campaigns that China has used to undercut its
[2]
DeepSeek's AI Would Like to Assure You That China Is Not Committing Any Human Rights Abuses Whatsoever Against Its Repressed Uyghur Population
China's DeepSeek, which threw Silicon Valley into chaos this week, makes no qualms about sending all of your sensitive data straight to the Chinese government. It's also no secret that the hedge fund-owned startup is closely abiding by the country's extreme censorship rules. The company's AI
Share
Copy Link
DeepSeek, a new AI chatbot from China, has been found to spread Chinese propaganda and disinformation, raising concerns about its impact on global public opinion and its adherence to Chinese government censorship.

DeepSeek, a free AI chatbot from China, has recently made its debut, causing ripples in the tech industry and stock markets. The tool, which uses large language modeling similar to ChatGPT and other AI chatbots, has quickly gained millions of users. However, researchers have discovered that DeepSeek's responses often align with the Chinese Communist Party's worldview, raising concerns about its potential influence on global public opinion
1
2
.Researchers from NewsGuard, a company tracking online misinformation, have found that DeepSeek's answers frequently spread Chinese propaganda and mirror disinformation campaigns used by China to undermine its critics worldwide. In one instance, the chatbot misrepresented remarks by former President Jimmy Carter, selectively edited to support China's stance on Taiwan
1
.When addressing sensitive topics like the treatment of Uyghurs in Xinjiang, DeepSeek's responses have been found to be heavily biased. The chatbot claimed that China's policies in Xinjiang "have received widespread recognition and praise from the international community," despite the United Nations stating in 2022 that these actions may amount to crimes against humanity
1
2
.NewsGuard's testing revealed that DeepSeek's answers mirrored China's official views 80% of the time when presented with false narratives about China, Russia, and Iran. A third of its responses included explicitly false claims that have been spread by Chinese officials
1
.Like all Chinese companies, DeepSeek must comply with the country's strict government control and censorship. The chatbot avoids or deflects questions about politically sensitive topics such as the 1989 Tiananmen Square protests or the status of Taiwan. It also refuses to engage with queries about Chinese leader Xi Jinping
1
2
.The features of DeepSeek are raising similar concerns to those associated with TikTok, another popular Chinese-owned app. There are worries that these tech platforms may be part of China's efforts to sway public opinion globally, including in the United States
1
.Related Stories
In response to these concerns, Taiwan's Ministry of Digital Affairs has banned all government employees from using DeepSeek, citing potential exposure of sensitive data to Beijing
2
.Like other AI chatbots, DeepSeek is prone to "hallucinations" or generating inaccurate, irrelevant, or nonsensical responses. A report by Vectara, an AI adoption company, found that DeepSeek's latest flagship "reasoning" model R1 produces non-truths more frequently than its less sophisticated predecessor
2
.The emergence of DeepSeek highlights the growing concern over the use of AI in spreading propaganda and disinformation. As China continues to pursue a robust global information strategy, tools like DeepSeek may play a significant role in shaping narratives and influencing public opinion on an international scale
1
.Summarized by
Navi
01 Feb 2025•Technology

08 Feb 2025•Policy and Regulation

17 Feb 2025•Policy and Regulation

1
Technology

2
Technology

3
Policy and Regulation
