YouTube will now auto-detect and label AI-generated realistic videos

YouTube is introducing automatic AI-generated video detection, clearer AI labels, and new personalised discovery feeds built around users' moods, interests, and prompts

YouTube is introducing a series of new updates focused on AI transparency and personalised content discovery on the platform. The company has announced more visible AI labels for videos, automatic AI-content detection tools and a new feature that allows users to create custom recommendation feeds based on their interests, moods, or prompts.

 

The updates come as AI-generated content becomes increasingly common on YouTube. While creators have been required to disclose realistic AI-generated content since 2024, YouTube said that it is now simplifying the process for both viewers and creators by making AI labels easier to notice and introducing systems that can automatically identify AI-generated videos when creators fail to disclose them.

 

 

YouTube’s AI labels: What’s new 

 

YouTube said that it is moving disclosure labels for photorealistic and meaningfully AI-generated or AI-edited videos to more visible areas of the platform.

 

According to the company, this will become the standard label format for realistic AI-generated content on YouTube. However, videos that are animated, unrealistic, or only slightly altered using AI will continue showing disclosures inside the expanded video description instead of the main label placement.

 

For long-form videos:

• The AI label will now appear directly below the video player
• The label will be placed above the description section

For Shorts:

• The disclosure will appear as an overlay directly on the video

 

How YouTube’s AI-generated video detection will work

 

Starting in May 2026, YouTube said that it will begin rolling out internal AI detection systems to identify realistic AI-generated content automatically. Creators will still be expected to manually disclose when they use realistic AI tools. But if a creator does not add the disclosure and YouTube’s systems detect significant AI-generated visuals, the platform will automatically apply an AI label to the video.

 

According to the company, creators will still have control over their content and can update disclosure settings through YouTube Studio if they believe a video was incorrectly flagged.

 

However, YouTube noted that AI labels will remain permanent in certain situations, including:

• Videos created using YouTube’s AI tools like Veo or Dream Screen
• Videos containing C2PA metadata showing they were fully AI-generated

YouTube also clarified that these labels will not affect monetisation or recommendations. YouTube said that the labels are meant for transparency. According to YouTube, the goal behind these changes is to help viewers better understand the kind of content they are watching as AI-generated videos become more realistic online.

 

  The company said that the labels are informational and are not meant to punish creators using AI tools responsibly.

 

 

Personalised custom feeds

 

Alongside the AI updates, YouTube has also announced a new custom discovery feature that lets users generate personalised recommendation feeds using prompts. Users can describe the kind of content they want to watch and YouTube will create a constantly refreshing feed based on that request. The generated feeds can also be pinned to the top of the Home page as custom chips for quicker access. The feature is designed to give users more direct control over recommendations instead of relying only on YouTube’s default algorithm.

 

Some example prompts shared by YouTube include:

• “Give me something different beyond my usual feed”
• “Help me unwind after work with guided meditations under 10 minutes”

How the custom feed feature works

 

Users can access the feature by:

• Tapping the “Your custom feed” chip on the Home page
• Entering a custom prompt
• Selecting suggested prompts if needed
• Editing prompts anytime to refresh the feed

Availability

 

The new custom feed feature is currently rolling out in English for signed-in users in the US on YouTube mobile apps and desktop. To use the feature, users must keep their YouTube watch history and search history turned on in account settings. YouTube also mentioned that users can send feedback directly if the recommendations do not match their prompts properly