There are two things needed for this to be successful,

1. Detection of AI voiceover. The article makes several references to photo realistic AI content but it's the voiceovers that are killing me.

2. Filtering options for viewers. It's not enough to be able to know if a video is generated. I don't want to see them, ever.