The core problem isn't the filter. The problem is how a lot of data with copyright status 'unknown' will get swept into the category 'copyrighted' even if most

The core problem isn't the filter. The problem is how a lot of data with copyright status 'unknown' will get swept into the category 'copyrighted' even if most of it probably isn't. Thinking about how to defeat the filter, while an interesting technical challenge, is irrelevant. No technological measure has made any dent in piracy. This was never a technology problem, it is a legal, social and moral one. Solving it with technology simply makes you a criminal, which like smoking pot will be ignored by the law until the law has a reason to look deeper into you. To wit, there is plenty of stuff today where the filter can't determine what it is. E.g. people playing old classical music, but some bigcorp has CDs with that same music. The problem is the filter errs on the side of blocking too much. This will also happen with the white noise if your scheme gets popular. Most of the last paragraph is incorrect, BTW. A creative step is always done by a human, never by a server, BTW. If the filter fails, the publisher of your video is on the hook for copyright infringement. The GDPR explicitly allows companies to gather data if there is a law requiring them to, so please don't open that can of worms - You don't want the Internet filter block access unless you install a browser plugin that scans your hard drive, for example.