From now on, videos generated by artificial intelligence that look credible will have to be flagged by their creators when uploaded to YouTube.

Specifically, they will have to select in this process whether the content shows someone doing or saying something that does not correspond to reality, alters a recording of a real place or event, or includes a scene that has not happened.

The platform thus follows in the footsteps of other environments such as TikTok or Facebook to try to tackle the risks of misinformation and manipulation that these pieces entail.

This danger is especially pronounced close to elections and the current year is historic because it has recorded at least 70 processes of this type in different parts of the world with hundreds of millions of citizens called to vote. Among them, presidential elections in the US or the European Parliament. And in fact, the capabilities of generative artificial intelligence also appear classified as threats in the 2023 Annual National Security Report that has just been approved by the Spanish National Security Council.

The labeling requirement is a key tool to reduce these risks but at the moment the experience is not showing good results. A recent study by the Stanford Internet Observatory on Facebook describes the prevalent use of manipulated or automatically generated images as a lever to increase page followers, despite the demands of the social network. In an undetermined percentage of cases they have a parodic or sarcastic spirit, but some people take them for real when they find them decontextualized in groups, which increases the problem.

In the process of uploading a video it will be necessary to select options in a menu that make clear the use of that technology if what it shows seems credible.

To avoid this situation, YouTube threatens offenders who may repeat failure to report the upload of this type of content with actions ranging from the removal of the video to the suspension of membership in the Partner program. The latter would cut off the advertising monetization of the affected users, which in turn would also reduce the possibility of advertisers appearing next to this type of creations and the eventual associated brand safety problems.

The new regulations do not apply in any case to those pieces in which it is evident that there is no intention of realistic replica or that do not have problematic consequences. The platform cites as examples the creation of animations or the automatic optimization of lighting or color settings. YouTube will also not force creators to reveal whether they have used that technology to generate scripts, content ideas or automatic subtitles.

https://dircomfidencial.com/marketing-digital/youtube-obliga-a-los-creadores-a-identificar-el-contenido-generado-por-inteligencia-artificial-20240320-1339/