YouTube has upgraded its measures for managing artificial intelligence content by using evaluation and labeling to notify users whether videos are AI-generated or human-created.
YouTube announced the deployment of an internal system to detect and label videos created with highly realistic AI, while enhancing label visibility and prominence so viewers receive transparent information.
This measure builds upon a policy enforced for over two years, under which creators on the platform were required to disclose their information via Creator Studio.
This development follows the recent launch of Gemini Omni, Google's AI model revealed at last week's Google I/O event, which demonstrated capabilities in producing high-quality videos and physics principles.
In cases where YouTube's automated system incorrectly labels a video, creators can request a correction; however, labels on videos created through AI tools cannot be removed, including videos embedding metadata compliant with the C2PA standard, an industry benchmark recently adopted by OpenAI.
Regarding on-screen display, labels will be relocated to more visible spots: for long-form videos, labels appear beneath the video player and above the video description, while on YouTube Shorts, labels overlay directly on the screen.
However, if an AI video is obviously AI-generated at a glance, the label will remain hidden within the description extension as before.
Concerning monetization, YouTube stated that these labels will not impact the video recommendation system or the creators' ability to earn revenue.