What Are Service Like Youtube Really Doing To Fight Copyright?

In this post I wanted to discuss what service like Youtube are really doing to prevent copyrighted content on their platform.

YouTube has a huge amount of copyrighted videos on their platform and I will be the first to admit I watch some of it.

My main vice is The First 48 I absolutely love this program if you are into real life crime documentaries check this out.

I watched the a couple of seasons on Amazon Prime the later ones and wanted to watch the first seasons but found out they are unavailable in our region.

This meant I had to try and find another way to watch it. I am subscribed to the Crime channel on Amazon but no luck on that.

I would like to state that like most people I am more than happy to pay for content I enjoy on a platform that I can easily stream from. I would happily pay around 5-9 pound per season.

There was nothing available the only place I could find to stream some of the content was Youtube.

I started watching a few episodes and they either had a really annoying backing track over the top or it was streaming a letter box space of content.

I posted a comment on why have the annoying backing music track and had a response that the episode will get removed from YouTube for copyright.

This was interesting for me having worked with AWS Rekognition tools for AI they are obviously running a scan on the audio to do some kind of shazam magic to look for copyright audio. Thats a good start then I started thinking how are they not using AI to also search through images extracted from the content to do comparisons this would surely eliminate a high proportion of copyrighted material.

The main reason is almost certain compute time and resources it would take a huge amount of resources to check every video via AWS or another similar tool to check every uploaded video.

I then started thinking does it really need to be that hard. You are always fighting against users trying to upload copyrighted content coming up with new ways to get around automated checks.

Obscuring the audio with annoying backing tracks
Only recording a certain space of the video

Sure it should be some kind of step by step service.

1. Check the audio for copyright
2. Check the video for image comparison copyright
3. Some kind of real life checks to determine if it is copyrighted.

Automating can only get you so far in certain situation you will always need a really person, we have not been taken over by robots yet.

Here is what I am looking to do take a video and try every hack I can think of then run it through AWS Artificial Intelligence tools and see what I can come up with.

I will post results below.

Its easier to think of ways to break something instead of prevent something.