YouTube updates AI video policies with an automatic detection system
The platform is expanding its AI transparency efforts with automatic labels for photorealistic AI videos and new personalised content feeds.
The platform says new disclosures are aimed at preventing viewers from being misled by generative AI content. (Image: Reuters) YouTube has announced a major update to how it handles AI-generated content, introducing automatic labels for videos that contain significant photorealistic AI elements, even when creators fail to disclose them manually.
The new system is designed to improve transparency around synthetic media as AI-generated videos become increasingly realistic and harder to distinguish from real footage. According to YouTube, the labels will specifically target content that has been “meaningfully altered or generated” using artificial intelligence in ways that could potentially mislead viewers.
Creators are still expected to manually disclose when realistic AI tools are used in their videos. However, YouTube says it can now automatically identify certain forms of AI-generated content and apply labels independently when necessary.
For long-form videos, the AI disclosure label will appear directly below the video player and above the description section. In YouTube Shorts, the label will be shown as an overlay on the video itself.
The platform clarified that the stricter labelling system mainly applies to photorealistic AI content. Videos featuring obviously animated, unrealistic, or lightly edited material will continue to use less prominent disclosures located within the expanded description section.
YouTube also noted that some disclosures cannot be removed manually. Videos created using the company’s own AI tools, including Veo and Dream Screen, or videos containing C2PA metadata indicating fully generative AI content, will permanently retain AI labels.
If creators believe a video has been incorrectly flagged as AI-generated, they can request adjustments through YouTube Studio.
Alongside the AI transparency update, the streaming company is also rolling out a new personalised content feed system. The feature allows users to generate custom video feeds based on specific moods, interests, or topics by entering prompts describing the kind of content they want to watch.
The personalised feed feature had been under testing since late 2025 and is now rolling out to signed-in users in the United States across YouTube’s mobile and desktop apps. The feature requires search and watch history to be enabled.
The move reflects the growing pressure on major platforms to improve disclosure standards around AI-generated media as synthetic video tools rapidly become more advanced and widely accessible.
Share
