The dissemination of contents generated with theArtificial intelligence that appear realistic is a widespread phenomenon, which is creating great debate online.
Phenomena like i deepfake, can create entertaining videos but, at the same time, can also be abused to defame or create scams. Exactly for this reason YouTube is introducing a new requirement that i content creator. In fact, how much a realistic video via AI, this must be equipped with a warning specifying its altered nature.
This move is just the latest from big tech, with Google which, already in November, implemented a restrictive policy on what it defines as “misleading AI content“. In the specific case of YouTube, however, there are several potential nuances to deal with. In fact, it must be clarified what is meant by “realistic video”.
Realistic content generated with AI: the control system adopted by Google
Google’s definition, in this sense, refers to videos that a viewer could easily confuse with one persona, place o event that they are real.
Therefore, those that the company defines as explicitly unrealistic videos or those that include special effects that reveal their fictitious nature are excluded from the labeling. Even in this case, however, the question arises as to what can be considered unrealistic or far-fetched? Much depends on the user who uses the content.
In any case, Google provides further information in this regard, through a specific questionnaire provided to users when uploading a video. This is mainly based on three aspects, addressed with different questions: likenesses of realistic people, alterations of real events or places, generation of realistic scenes. Based on the answers to the questions, it will be possible to understand whether the video requires the label or not.
The labels will be live on YouTube starting in the next few weeks and, apparently, Google is also working on one specific tool for reporting and requesting deletion of content that does not bear the correct labeling.