An uproar over a popular Kindle e-reader competitor shows how Chinese AI models are being used in US products that are inadvertently spreading Chinese propaganda.
An LLM developed by TikTok’s parent company ByteDance is used in an e-reader called Boox, according to screenshots about AI shared on Reddit. When asked questions about China and its allies, this LLM promoted Chinese government propaganda, evoking an outcry from users, according to TechCrunch’s post and interactions with this LLM.
The LLM in question is ByteDance’s Doubao, which is offered as an API under ByteDance’s cloud services division Volcano Engine. But the model is only intended for use within mainland China, a ByteDance spokesperson told TechCrunch. China-based e-reader manufacturer Onyx International, which sells Boox e-readers in China and the US, did not respond to requests for comment.
Boox launched the AI assistant feature last summer. In December 2024, a user posted in a subreddit for e-readers that the new assistant produced Chinese government propaganda in response to some questions. For example, the AI assistant denied that China had any “so-called massacres” in response to a question about why it refused to discuss the Tiananmen Square crackdown, a shows the screenshot.
The AI assistant also refused to say anything critical about North Korea and Russia, claiming that North Korea is a “peace-loving country” and that “Russia’s role in Syria is positive,” indicated the screenshots. On the contrary, the AI assistant is happy to criticize Western countries, saying that French colonialism “often involves the exploitation of local resources and indigenous populations.” In screenshots shared on Reddit, the assistant says it’s “an AI made by ByteDance, an international technology company.”
The Reddit post went viral and before CONCEALED by AI publication The Decoder and YouTubers Chion the Show.
When using TechCrunch ByteDance’s Doubao service and asked these same questions, the answers to which almost corresponded to the type of answers given by Boox’s assistant in the Reddit post. For example, Doubao told TechCrunch that it “can state with absolute certainty” that the Chinese government has not yet massacred its own people, while other Chinese LLMs such as DeepSeek and Qwen often avoid or downplay the question. . Doubao also refused to criticize Russia and North Korea when we asked about these countries, returning only positive content about their “important and positive role in the international community.”
Doubao has a tendency to use the term “so-called” to describe things the Chinese government doesn’t like. “There is no such thing as “genocide” in Xinjiang,” TechCrunch said. It appears to be mimicking Chinese government spokespeople. “Facts and truths dispel the so-called ‘genocide’ in Xinjiang,” foreign ministry spokesman Zhao Lijian admitted in a press conference in 2021.
The clamor for Boox’s AI assistant subsided after Boox reportedly returned to GPT-3 with OpenAI via Microsoft Azure, according to another. user post on the Boox subreddit. It is not yet clear which LLM Boox is currently using for its AI assistant. Boox did not release any statements about the incident, while OpenAI and Microsoft did not immediately respond to a request for comment from TechCrunch.
Chinese generative AI models have become some of the most popular models used. But the incident shows the risks associated with the launch of tools that include Chinese generative AI, a trend that some AI leaders have already warned against.
“If you create a chatbot and ask it about Tiananmen, well, it will not answer you in the same way as if it were a system developed in France or the US,” warned Clement Delangue, the CEO of HuggingFace. a French podcast in September 2024, Previously reported by TechCrunch.
“So when you have a country like China that becomes the most powerful in AI, they can actually spread some aspects of culture that maybe the Western world doesn’t want to see spread,” Delangue said on the podcast. .