On September 15, WeChat's AI assistant Xiaoxiao became a hot topic on Sina Weibo due to privacy issues. WeChat responded by clarifying that claims such as "Xiaoxiao directly reads chat records and spies on all private information" are all false.
According to the introduction, Xiaoxiao is a native AI assistant developed by the WeChat team, currently still in a small-scale beta test phase. It supports text and voice interaction, and can provide services such as weather, express delivery, and takeout inquiries. It can also be proactively triggered in scenarios such as Moments, official accounts, chats, and video numbers to perform operations such as content summarization.
WeChat stated that the information Xiaoxiao can access will not exceed the scope of information that users themselves can view. It will not read deleted Moments posts, and content recommendations will not exceed the user's own information access permissions. In chat scenarios, Xiaoxiao will only temporarily call up specific chat records when the user actively asks a question, and after processing, the original chat records will be handled in a "use-and-discard" manner without being retained.
This response focuses on the privacy boundary issues arising from AI assistants deeply integrating into social scenarios. As AI gradually moves from independent applications into frequent scenarios such as instant messaging and content consumption, the scope of data access, user authorization mechanisms, and the way data is retained after processing have become important factors affecting the large-scale application of AI products.
