Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
An uproar with the e-reader, a popular Kindle competitor, showed how the use of Chinese AI models in US products could unwittingly spread Chinese propaganda.
A large language model (LLM) developed by TikTok’s parent company ByteDance used an e-reader called Boox, according to Screenshot about AI Shared on Reddit. When asked questions about China and its allies, the LLM ran Chinese government propaganda, prompting outrage from users, according to the Post and TechCrunch’s interactions with the LLM.
The LLMT in question was ByteDance’s Duba, which is offered as an API under ByteDance’s cloud services division Vulcano Engine. But the model is only for use in mainland China, a ByteDance spokesperson told TechCrunch. The e-reader’s China-based manufacturer, Onyx International, which sells box e-readers in China and the United States, did not respond to requests for comment.
Boox introduced AI assistant feature last summer. In December 2024, a user has been posted In a subreddit for e-readers, the new assistant is creating a Chinese government campaign to answer some questions. For example, when asked why it refused to discuss the Tiananmen Square crackdown, the AI assistant denied there was a “so-called genocide” in China. As the screenshot shows.
The AI assistant also refused to say anything critical about North Korea and Russia, claiming that North Korea is a “peaceful country” and that “Russia’s role in Syria has been positive,” the screenshots show. In contrast, the AI assistant was happy to criticize Western countries, noting that French colonialism “often involved the exploitation of local resources and local populations.” In screenshots shared on Reddit, the assistant says it’s “an AI developed by ByteDance, an international technology company.”
Reddit post has gone viral and was covered AI Publication by The Decoder and YouTubers China Show.
When using TechCrunch ByteDance’s diving services and asked it a similar question, its answers closely matching the responses given by Boox’s assistant in the Reddit post. For example, Doubao told TechCrunch that “it can be said with absolute certainty” that the Chinese government has never committed genocide against its own people, while other Chinese LLMs such as DeepSeek and Qwen generally avoid or downplay the question. Doba also declined to criticize Russia and North Korea when we asked about those countries, only returning to positive content about their “important and positive role in the international community.”
Dobao has a penchant for using the word “so-called” to describe things the Chinese government doesn’t like. “There is no so-called ‘genocide’ in Xinjiang,” it told TechCrunch. It appears to mimic Chinese government spokespeople. “Facts and truths have exposed the so-called ‘genocide’ in Xinjiang,” Foreign Ministry spokesman Zhao Lijian claimed at an event. Press conference in 2021.
The furor over Boox’s AI assistant has subsided after Boox moved back to OpenAI’s GPT-3 via Microsoft Azure, according to another User posts in the Books subreddit. It is not yet clear which LLM Boox currently uses for its AI assistant. Boox has not released a statement about the incident, while OpenAI and Microsoft did not immediately respond to requests for comment from TechCrunch.
The Chinese have become something of a generative AI model The most popular model used. But the incident shows the risks involved in introducing tools that incorporate Chinese generative AI, a trend some AI leaders have already warned about.
“If you build a chatbot and ask it a question about Tiananmen, well, it’s not going to give you the same answer as a system built in France or the US,” warns Hugging Face CEO Clément Delangue. In a French podcast in September 2024, TechCrunch previously reported.
“So if you have a country like China that becomes the most powerful in AI, actually they’re going to be able to spread some cultural aspects that maybe the Western world doesn’t want to spread,” Delangue said on the podcast.
