I hope they broaden the criteria to include AI-voiced videos. I am often suggested shorts that seem interesting from the thumbnail and title but as soon as you click it's an AI voiced and scripted video using stock footage that is just pointless and awful.

skytube lets you ban chans, idk how anyone finds anything common and useful without it, the autogenerated yt spam has made the real search interface pointless.

What a great idea! Soon there could be "AI video detection tools" that uses "AI" to determine if the video uses "AI", and then some tools could pop up telling how to bypass the " AI detection" using their state-of-art AI video generation tools.

If anyone thought jobs were going to disappear, they need to read this.

So most videos from smartphones have to be labelled now because they are almost all enhanced in some way.

Dupe: https://news.ycombinator.com/item?id=38269656

crosspost https://news.ycombinator.com/item?id=38269656

Really curious how they are going to enforce this at scale.

Insufficiently, with AI. It's bullshit all the way down.

"We used the AI to check the AI"

[dupe]

More discussion here: https://news.ycombinator.com/item?id=38269656

Instruction videos that have more than 30sec of introductory material should also be labelled.

Even if part of the script was written by AI? I doubt they can prove that

I don't think that's what they are going for here:

> For example, this could be an AI-generated video that realistically depicts an event that never happened, or content showing someone saying or doing something they didn't actually do.

Getting ChatGPT to help you write a script that you then deliver seems to me like it wouldn't match those.

And how are they going to enforce this? Even worse content ID?