YouTube will begin automatically detecting and labelling videos that contain significant photorealistic Artificial Intelligence-generated content, replacing a system that has relied on creator disclosure since it launched in 2024. The automatic labels will roll out gradually, starting in May 2026. The company said it will use internal signals to identify Artificial Intelligence-generated material, though it did not detail the technical methods behind the system.
The platform is also making labels much more visible. For long-form videos, labels will appear directly below the video player instead of being buried in the expanded description. For Shorts, the label will appear as an overlay on the video itself. Previously, prominent labels were mainly reserved for sensitive subjects such as health, news, elections, or finance, while other disclosures were less noticeable. That distinction is being removed, and every Artificial Intelligence-labelled video will now carry a visible marker regardless of topic.
Creators will still be able to update their disclosure if they believe a video has been incorrectly flagged. But YouTube is making labels permanent when content is created with YouTube’s own Artificial Intelligence tools, including Veo, Gemini Omni, and Dream Screen, and when C2PA metadata shows that the content is fully Artificial Intelligence-generated. YouTube’s detection system will also read both C2PA metadata and SynthID watermarks as part of the identification process. C2PA was founded in 2021 and has since grown to more than 6,000 member organisations. SynthID has been applied to more than 100 billion Artificial Intelligence-generated images and videos to date.
The labelling push is arriving alongside wider deepfake protections. On 16 May 2026, YouTube extended its deepfake complaint system to all adults aged 18 and over. Previously, only public figures, including creators above a certain following threshold, celebrities, politicians, and journalists, could request removal of Artificial Intelligence-generated content using their likeness. The current system covers face-based deepfakes, while voice cloning detection is expected later in 2026.
YouTube said labelled videos will not be downgraded in recommendations or lose monetisation, describing the change as a transparency measure rather than a penalty. The move comes ahead of the European Commission’s Artificial Intelligence Act transparency obligations, which take effect in August 2026. Across the industry, platforms including Meta and TikTok have introduced their own approaches to Artificial Intelligence labelling, but YouTube’s shift reflects a broader recognition that voluntary disclosure has been unreliable. The company is betting that clearer provenance signals and automated detection will make synthetic media easier to identify as Artificial Intelligence creation tools become more common on the platform.
