YouTube announced major updates to its content handling policies created by artificial intelligence (AI). This adds a system that automatically detects and labels reliable AI videos.
According to YouTube, the move is aimed at increasing transparency, as AI video creation tools are becoming increasingly powerful and difficult to distinguish from real images.
The new system will focus on videos that have been altered or intentionally created by AI in a way that could mislead viewers. Content that falls into this category will be labeled even if the author has not self-declared the use of AI.
YouTube previously required creators to proactively disclose whether their videos use AI to simulate images, audio, or realistic elements. But the company said it can now automatically detect some types of AI content and proactively label it if needed.
For longer videos, a notice about the use of AI will appear just below the video player and above the description. YouTube Shorts, on the other hand, displays the AI label directly as the cover of the video.
YouTube said the stricter labeling system will primarily be applied to authentic AI videos, such as real-life simulated scenes or real-life events. Animated videos, unrealistic content, or simple edits continue to use lightweight notification forms with extended descriptions.
Some AI labels cannot be deleted manually. Videos created using YouTube’s AI tools, such as Veo and Dream Screen, or videos that include C2PA metadata confirming content created by AI, will always have an AI-recognized label.
If you believe a video is incorrectly labeled, creators can submit a review request through YouTube Studio.
In addition to the AI policy changes, YouTube also introduced a new personalized content sourcing system. This feature allows users to create a list of videos according to their taste by entering a description of the content they want to watch.
For example, users can ask the video suggestion system about their mood, favorite topics, or specific types of content. YouTube uses your viewing and search history to build feeds that are more relevant to you.
The company says the feature has been in testing since late 2025 and is currently being rolled out for logged-in users in the United States on both the mobile application and the computer version. To use it, you need to turn on search history and video viewing history.
