what exactly falls under this category?
Well, if a video makes it seem like a real person is saying or doing something they didn't actually do, if it alters footage of real events or places, or even if it depicts a scene that looks real but didn't really happen, creators need to check the box and let viewers know. For instance, if you show a fake tornado heading towards a real town or use deepfake voices to make a real person narrate your video, you'd need to disclose that.
However, not all enhancements require disclosure. Things like beauty filters, background blur effects, or content that's clearly unrealistic, like animation, won't need to be flagged.
Back in November, YouTube outlined its policy regarding AI-generated content, dividing it into strict rules for protecting music labels and artists and looser guidelines for others. For instance, if an artist doesn't like a deepfake music video, their label can request its removal. As part of these rules, YouTube stated that creators would have to disclose AI-generated content, but they hadn't explained how until now. If you're an average person featured in a deepfake video on YouTube, getting it taken down might be tougher, as you'd have to fill out a privacy request form for YouTube to review.
what is the new update?
Today's update from YouTube didn't delve into specifics about this process, stating only that they are "working towards an updated privacy process."
Similar to other platforms, YouTube's new feature operates on the honor system, relying on creators to be truthful about the content in their videos. Although YouTube mentioned investing in tools to detect AI-generated content, the accuracy of AI detection software historically has been questionable.
YouTube also mentioned the possibility of adding AI disclosure to videos themselves, particularly if the content has the potential to confuse or mislead viewers. Additionally, more prominent labels will be added to videos covering sensitive topics like health, elections, and finance.