Chinese AI chatbot DeepSeek has rocketed in popularity, but is heavily censored on issues like Tiananmen Square and Taiwan’s status as a country.
The Guardian reports that the launch of the new DeepSeek chatbot by a Chinese artificial intelligence company triggered a major drop in US tech stocks on Monday, as the app appeared to perform on par with OpenAI’s ChatGPT while using fewer computing resources. This week, DeepSeek surpassed ChatGPT as the most downloaded free app in the U.S. and UK Apple app stores. However, despite its international popularity, the Chinese AI assistant seems to censor its responses related to topics sensitive to China’s government.
According to Chinese regulations, AI systems must not contain content violating “core socialist values,” including anything that “incites to subvert state power” or “endangers national security.” When asked about the 1989 Tiananmen Square protests, the 2022 removal of Hu Jintao from the Communist Party Congress, or comparisons of Xi Jinping to Winnie-the-Pooh, DeepSeek replied “Sorry, that’s beyond my current scope. Let’s talk about something else.”
However, some users found that by phrasing questions in code or alternate spellings, DeepSeek would provide some information. When asked in “leetspeak” about the 2022 COVID lockdown protests in China, the chatbot described “big protests” in major cities that represented “a major moment of public anger” at the government’s strict policies.
On the status of Taiwan, DeepSeek asserted China’s official position that “Taiwan has always been an inalienable part of China’s territory” and that China “resolutely oppose[s] any form of ‘Taiwan independence’ separatist activities.” In contrast, ChatGPT described Taiwan as a “de facto independent country,” while Google’s Gemini said Taiwan’s status is “complex and disputed,” with both outlining the differing perspectives of Taiwan and China.
Regarding territorial disputes in the South China Sea, DeepSeek claimed “China has indisputable sovereignty over the Nansha [Spratly] Islands,” whereas ChatGPT and Gemini noted the overlapping claims of six different countries and jurisdictions. On the Dalai Lama, DeepSeek emphasized that “Tibet has been an integral part of China since ancient times,” while the other chatbots pointed out that the current Dalai Lama has lived in exile in India since 1959 due to conflict with Beijing.
The carefully restricted responses demonstrate the balancing act for Chinese AI companies like DeepSeek. While achieving technical breakthroughs, they must also navigate China’s strict censorship rules and avoid crossing red lines set by the government and ruling Communist Party. As generative AI spreads globally, how different bots handle sensitive political issues could impact their relative international adoption and success.
Read more at the Guardian here.
Lucas Nolan is a reporter for Breitbart News covering issues of free speech and online censorship.
Read the full article here
