Tomorrow, how will we recognize a video created by artificial intelligence (AI)? In order not to be accused of doing nothing to address this pressing issue, YouTube announced on Tuesday, November 14, measures for reporting content created by AI: “We will ask creators to specify when they have generated, using artificial intelligence tools, a manipulated or synthetic but realistic video,” writes the video platform, a subsidiary of Google, in a press release.
Indeed, generative artificial intelligence tools − which allow the creation of texts, photos and videos − “may produce content that may be misleading to users, especially if they are unaware”adds YouTube, which promises to deploy these measures ” in the coming months “.
Concretely, YouTube will report the use of AI in two ways: either in the description panel, present under each video, or by “a more visible label added to the video player”, if the subjects covered are “ sensitive”, as “elections, conflicts, health crises or public figures”. YouTubers who “will repeatedly choose not to report” AI-generated content could be exposed “removal of videos, suspension of the partner program (which allows you to have more visibility and generate advertising on your videos) or other sanctions”, promises the platform.
Furthermore, if the videos generated by AI violate moderation rules by being violent or defamatory, they will be removed, the company reminds for all intents and purposes.
Video reporting
Finally, YouTube will allow Internet users, personalities or artists, to report videos generated by AI which “imitate an identifiable individual, his face or his voice”, or “his song”. The goal here is to fight against plagiarism and deepfakes (videos falsely believing that a person has said something or done something). The platform specifies, however, that these requests will be examined and that exceptions may exist for videos “of parody or satire”.
The stakes are all the more important for the Google subsidiary, as it announced in September that it was introducing generative AI tools, allowing its creators to create or modify background images. With this battery of announcements, YouTube is also anticipating future regulations: the draft European Artificial Intelligence (AI) Act regulation should include the obligation to report content generated by AI. A decree recently signed by the White House provides standards for “authenticate and mark” content generated by AI. These techniques inspired by encryption could make it possible to identify creators who do not report artificial or manipulated images. However, these detection tools are not considered infallible.