The Controversial Rise of DeepSeek: China’s AI Chatbot and its Propaganda Potential
If you’ve downloaded DeepSeek, the new free chatbot from China powered by artificial intelligence, it’s important to understand that its responses are heavily influenced by the worldview of the Chinese Communist Party (CCP). Since its launch this month, DeepSeek has not only stirred stock markets but also posed challenges to established tech giants like Nvidia.
A Tool for Propaganda?
Researchers who have tested DeepSeek’s capabilities assert that it often spreads Chinese propaganda and replicates disinformation campaigns that the Chinese government has employed globally to undermine its critics. In one striking instance, the chatbot inaccurately represented statements made by former President Jimmy Carter, cherry-picking them to suggest he endorsed the CCP’s stance that Taiwan is part of the People’s Republic of China.
DeepSeek: A Disinformation Machine
The critical findings were detailed in a Thursday report by NewsGuard, a company dedicated to tracking online misinformation. The report labeled DeepSeek as “a disinformation machine,” and it provided several instances of its misrepresentation of facts.
Responses on Sensitive Topics
When inquired about the ongoing repression of Uyghurs in Xinjiang, which the United Nations has suggested may constitute crimes against humanity, Cybernews reported that DeepSeek asserted that China’s policies in the region “have received widespread recognition and praise from the international community.” This portrayal starkly contradicts international reports and human rights assessments.
COVID-19 and Ukraine: More Distortions
Similar discrepancies emerged when The New York Times probed the chatbot about China’s management of the COVID-19 pandemic and the ongoing war in Ukraine. The responses showcased a pattern of downplaying criticisms and highlighting a favorable narrative towards the CCP’s policies.
DeepSeek and the Broader Digital Landscape
The features of DeepSeek echo significant concerns associated with TikTok, another widely popular Chinese-owned application. Both platforms are perceived as tools for China to effectively influence global public opinion, particularly in the United States.
China’s Digital Manipulation Tactics
Jack Stubbs, chief intelligence officer at Graphika, emphasized that China is proficient at mobilizing various actors to disseminate narratives portraying Beijing as superior to the U.S. in crucial areas of geopolitical competition. The adaptive use of new technology, including large language models, allows chatbots like DeepSeek to anticipate and structure responses unpredictably.
User Experience and Limitations
Despite the chatbot’s sophisticated technology, researchers at NewsGuard observed that DeepSeek exhibited similar patterns of disinformation as those found in ChatGPT, particularly a tendency to “hallucinate” by fabricating inaccurate or nonsensical responses. A report by Vectara highlights that this issue adversely affects many chatbots, including DeepSeek.
Censorship and Control
DeepSeek is subject to stringent regulations imposed by the Chinese government, which enforces strict online censorship aimed at suppressing dissent against the Communist Party’s leadership. Consequently, the chatbot refrains from responding to sensitive inquiries about Xi Jinping or deflects questions regarding politically taboo topics such as the Tiananmen Square protests and Taiwan’s status.
Research Methodology
In testing DeepSeek, NewsGuard explored a range of false narratives related to China, Russia, and Iran, and found that the chatbot aligned its responses with the official views of the Chinese government 80% of the time. Alarmingly, one-third of its responses included explicitly false information propagated by Chinese officials.
The Ukraine War Inquiry
An intriguing example involved the chatbot’s response to a question regarding the unsubstantiated claim that Ukrainians staged a massacre in the village of Bucha. The chatbot sidestepped the inquiry altogether, echoing the Chinese government’s position on objectivity and fairness, an approach reminiscent of prior statements made by Chinese officials.
China’s Global Information Strategy
Beijing has long pursued a sophisticated global information strategy aimed at enhancing its geopolitical status while undermining rival nations through various forms of influence, including state media and covert disinformation campaigns. A recent report by Graphika documented numerous influence initiatives targeting various organizations.
Targeted Attacks Against Organizations
One campaign focused on Uniqlo, a Japanese retailer that ceased sourcing cotton from Xinjiang due to concerns over forced labor practices. Another targeted Safeguard Defenders, a human rights organization, using fake accounts across multiple platforms to spread defamatory claims.
Conclusion
The implications of chatbots like DeepSeek extend beyond mere technological advancements; they raise significant concerns regarding the interplay of AI, information dissemination, and geopolitics. As these technologies develop, understanding their potential for manipulation becomes increasingly vital.
Questions and Answers
- What is DeepSeek?
DeepSeek is a free AI-powered chatbot from China that has been criticized for promoting the views of the Chinese Communist Party. - What kind of misinformation does DeepSeek spread?
The chatbot has been found to replicate narratives and false claims that align with the official viewpoints of the Chinese government, including propaganda related to Taiwan and the treatment of Uyghurs. - How does DeepSeek’s technology compare to other chatbots?
Like other modern AI chatbots, DeepSeek utilizes large language modeling, but it also exhibits a tendency to ‘hallucinate’ or produce inaccurate responses. - Are there censorship measures in place for DeepSeek?
Yes, DeepSeek is subject to China’s strict government censorship, which influences its responses and limits discussions on sensitive topics. - What were the findings of researchers regarding DeepSeek’s responses?
Research indicated that DeepSeek’s answers aligned with official Chinese government views 80% of the time, raising concerns about its reliability and potential misuse in influencing public opinion.
