As artificial intelligence (AI) continues to advance, companies and platforms are grappling with how to navigate the increasing prevalence of AI-generated content.
Read More: Telegram Gets New Business Functionalities, Including Greeting Messages, Quick Replies
One of the central concerns is ensuring transparency by clearly identifying content created by AI models, whether it be in the form of photos, videos, or audio recordings. In response to this challenge, YouTube has announced its latest guidelines for labelling videos produced with artificial intelligence.
Effective immediately, YouTube will mandate that any user uploading a realistic-looking video, created using altered or synthetic media, including generative AI, must label it accordingly. The platform defines realistic content as material that viewers could easily mistake for genuine footage of a person, event, or location.
Creators utilising synthetic versions of real voices to narrate videos or substituting faces in footage will be required to include a label. Similarly, alterations to footage of actual events or places, such as modifying cityscapes or simulating fires, will necessitate disclosure.
YouTube has warned that failure to comply with these regulations may result in penalties, particularly for persistent offenders who neglect to label content as required.
Read More: YouTube revamps TV app to deliver a ‘rich experience’
The platform aims to gradually implement these labels across its platform, beginning with the mobile app and subsequently extending to desktop and television interfaces. Labels will primarily appear in expanded video descriptions, indicating that the content includes altered or synthetic elements, with a note specifying significant editing or digital generation of sound or visuals.
In cases involving sensitive topics like news, elections, finance, and health, YouTube plans to prominently display labels directly on the video player to enhance visibility.
However, creators will not be obligated to label content if they solely utilise generative AI for tasks such as script creation, generating video ideas, or automatically generating captions. Furthermore, labels will not be required for content deemed “clearly unrealistic” or if alterations are deemed inconsequential, such as colour adjustments or basic special effects like background blurring. Similarly, applying lighting filters, beauty filters, or other enhancements will not necessitate the use of altered content labels.
Read More: How to set and change ‘Home’ and ‘Work’ location in Google Maps
YouTube is also in the process of developing an updated procedure for handling takedown requests related to synthetic or altered content featuring identifiable individuals. Further details regarding this revised process will be shared by the platform in due course.
