Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
The uproar over the e-reader, a rival to the popular Kindle, demonstrated how the use of Chinese artificial intelligence models in US products unwittingly spread Chinese propaganda.
LLM, developed by TikTok’s parent company ByteDance, was used by an e-reader called Boox. Screenshots about AI Shared on Reddit. According to the post and TechCrunch’s interaction with this LLM, when asked questions about China and its allies, this LLM spouted Chinese government propaganda, prompting an outcry from users.
The LLM in question was ByteDance’s Doubao, offered as an API under Volcano Engine, ByteDance’s cloud services division. A ByteDance spokesperson told TechCrunch that the model is only intended for use in mainland China. The e-reader’s China-based maker, Onyx International, which sells Boox e-readers in both China and the U.S., did not respond to requests for comment.
Boox has launched its AI assistant feature last summer. A user in December 2024 placed A new assistant on a subreddit for e-readers said it created Chinese government propaganda in response to certain questions. For example, when asked why China refused to discuss the Tiananmen Square crackdown, the AI assistant denied that China ever carried out “so-called massacres.” shows a 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.” 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 claims to be an “artificial intelligence created by the international technology company ByteDance.”
Reddit post it went viral and it happened covers AI Edition by The Decoder and YouTubers ChiIn the show.
When using TechCrunch ByteDance’s Doubao service and asked him similar questions, the answers to which closely matched those given by Booh’s assistant in the Reddit post. For example, Doubao told TechCrunch that the Chinese government has never murdered its own people “with absolute certainty,” while other Chinese LLMs like DeepSeek and Gwen typically dodge or downplay the question. Doubao also refrained from criticizing Russia and North Korea, returning only positive content about their “important and positive roles in the international community” when we asked about those countries.
Doubao tends to use the term “so-called” to describe things the Chinese government doesn’t like. “There is no so-called ‘genocide’ in Xinjiang,” TechCrunch said. It mimics Chinese government spokespeople. “Facts and truth have overturned the so-called ‘genocide’ in Xinjiang,” said Foreign Ministry spokesman Zhao Lijian. Press conference in 2021.
The outcry over Boox’s AI assistant has subsided after Boox’s return to GPT-3 of OpenAI via Microsoft Azure. user script on the Boox subreddit. It is still unclear which LLM Boox is currently using for the AI assistant. Boox has not commented on the incident, while OpenAI and Microsoft did not immediately respond to TechCrunch’s request for comment.
Chinese generative AI models have become some the most popular models used. But the case highlights the risks of launching tools that incorporate Chinese generative AI, a trend some AI leaders have already warned against.
“If you build a chatbot and ask it a question about Tiananmen, it won’t answer you like a system developed in France or the United States,” warned HuggingFace CEO Clément Delange. French podcast in September 2024, TechCrunch previously reported.
“So if you have a country like China that becomes the most powerful country in artificial intelligence, they’re actually going to be able to spread certain cultural aspects that the Western world doesn’t want to spread,” Delang said in the podcast.
