YouTube introduces automatic AI labeling for photorealistic content and customizable feeds

At a glance:

• YouTube will automatically label videos with significant photorealistic AI use, even if creators don't disclose it.
• Creators can still manually update disclosures in YouTube Studio, but labels from YouTube's own AI tools (Veo, Dream Screen) or C2PA metadata are permanent.
• A new customizable content feed feature is rolling out to U.S. users on mobile and desktop, allowing personalized prompts.

YouTube is stepping up its efforts to transparently flag AI-generated content with a new automatic detection and labeling system. Starting this week, the platform will proactively identify and label videos that feature "significant photorealistic AI use," addressing growing concerns about synthetic media blurring the lines between real and artificial content. This move comes as regulators and users increasingly demand clarity on AI-generated material, especially in an era where deepfakes and hyper-realistic AI videos can easily mislead audiences.

The automatic labeling system targets content that could "fool people" due to its photorealistic quality, distinguishing it from clearly unrealistic or animated material. For long-form videos, the AI label will appear just below the video player and above the description, while Shorts will display an overlay label directly on the video. This approach ensures visibility without disrupting the viewing experience for less deceptive content, which remains relegated to expanded descriptions as before.

Creators retain some control over the process. Despite the automation, they are still required to manually disclose AI use in their videos. If a video is incorrectly flagged, creators can adjust the disclosure status through YouTube Studio. However, disclosures tied to YouTube's proprietary AI tools—such as Veo and Dream Screen—or content with C2PA (Coalition for Content Provenance and Authenticity) metadata indicating full generative AI will remain permanently labeled. This distinction underscores YouTube's push to standardize AI content identification across its ecosystem.

The update also introduces a customizable content feed, allowing users to generate personalized video streams based on specific interests, moods, or topics. By typing a custom prompt, viewers can curate feeds tailored to their preferences, a feature that has been in testing since November 2023. Now rolling out to signed-in U.S. users on both mobile and desktop platforms, the tool requires YouTube search and watch history to be enabled. This aligns with broader trends in algorithmic personalization, though it raises questions about filter bubbles and user agency in content discovery.

YouTube's AI labeling initiative follows similar moves by platforms like Meta and TikTok, which have also implemented AI content detection and labeling. These measures are part of a larger industry response to regulatory pressure, such as the EU's Digital Services Act, which mandates transparency around AI-generated content. While YouTube's approach focuses on photorealistic content, the platform may expand its criteria as AI capabilities evolve and public expectations shift.

The customizable feed feature, meanwhile, reflects YouTube's ongoing competition with TikTok and Instagram Reels, where personalized content curation plays a central role. By giving users more control over their feeds, YouTube aims to enhance engagement while maintaining its position as a go-to platform for both creators and viewers. However, the reliance on search and watch history for the feature to function highlights the platform's continued emphasis on data-driven recommendations, a strategy that has drawn scrutiny over privacy and algorithmic bias concerns.

Looking ahead, YouTube's dual focus on AI transparency and user customization signals a balancing act between innovation and accountability. As AI-generated content becomes more sophisticated, the platform's labeling system will likely face tests in accuracy and consistency. Similarly, the customizable feed's success will depend on how well it adapts to user feedback and avoids reinforcing existing content silos. Both features represent incremental but meaningful steps in YouTube's broader strategy to navigate the evolving digital media landscape.