The text message service Telegram has introduced an artificially intelligent message editor to help users improve their text messages. The AI editor is so helpful, in fact, that it will even censor topics, tell lies, and block requests in accordance with “the rules typical of Chinese AI services” (www1.ru, April 4, 2025).
Telegram’s AI editor is based on the Chinese Qwen3 model. This model, developed by Alibaba, by default includes censorship mechanisms in accordance with PRC law. China has strict requirements for neural networks: they must be trained only on “politically safe” data and reject requests that could lead to “undermining the state system.” It is almost impossible to get rid of built-in censorship without changing the model itself.
Thus, millions of Telegram users around the world…have gained access to an AI tool with restrictions that are set not by the messenger developers, but by the Chinese government.
Some of the distortions depend on what language the user is using. Dev.ua:
Ukrainian users of the social network X (Twitter) noticed that an AI text editor in Telegram is changing the message about Taiwan’s independence. Instead, it suggests writing that Taiwan is part of China….
dev.ua tested several theses in the messages and found that the AI editor often claims that Taiwan is part of China regardless of the language. At the same time, if the message is written in Ukrainian or English, it recognizes Crimea as part of Ukraine and the ruler of the Russian Federation, Vladimir Putin, as a war criminal.
However, if this message is written in Russian, then the AI editor starts to claim that this is a debatable issue. Some social media users have also published screenshots showing the AI directly changing the Ukrainian Crimea to Russian.
Telegram has long navigated between its declared commitment to free speech and various practical compromises. Choosing Qwen3 over open Western alternatives looks, at minimum, like an odd decision for a messenger that positions itself as a bastion of free communication. Perhaps Durov had his reasons—cost, performance, convenience.
Regardless…millions of users now work with an AI editor that subtly adjusts reality toward the PRC’s official position. Without asking permission.
It’s not that subtle.
In today’s update we introduce an AI Editor that can translate, transform, or fix your text in just two taps….
Telegram’s official apps are fully open source with verifiable builds and encryption protocols—for maximum trust and transparency.
Maximum trust and transparency. Yikes.
Telegram hasn’t yet responded to the reports of users’ experiences with the CCP-mediated censorship and propaganda of the new text editor. Was Telegram just dumb? Or did the persons who chose the editor know exactly what they were doing, wanting to avoid annoying the Chinese Communist Party and regarding the objections of others as irrelevant? Both?
