YouTube now lets people report deepfakes of their face or voice, but the manual review takes up to a week while bad actors churn out fakes with free tools. Categorizing it as a privacy violation instead of a content violation also limits how hard YouTube can come down on repeat offenders.
Chapters
YouTube has announced a new feature where people can report if a video is showing a likeness of them, of their voice or image, that has been generated by artificial intelligence, essentially a deepfake. And this is a great thing. It's a great step forward because we are seeing more abuse of these deepfakes being created and published to social media sites. But there is a downside. Once it's reported, it may take up to a week and YouTube has to go out and manually look and make a decision if a deepfake is inappropriate. And that can create a problem. It also means that it's a lot of work to be done. And malicious users can create these deepfakes very easily with free and automated tools. And that just creates a bigger problem. So, I don't see how YouTube can keep up with the demand. And a violation of this is now being considered a privacy violation versus a content violation. And the difference there is YouTube won't take direct action in suspending the account for a privacy violation like it can and does for content violations. So this is an important step forward. I don't see it as being scalable, but I think it starts to send a message and we start moving down the path of getting better tools and processes to help victims of deepfakes. We need these. We need to protect people's privacy and dignity. So this is a good first step, but this isn't the end of the story.
