> Youtube does state that explicit materials are allowed for educational content, but if their copyright machines can't even distinguish white noise from copyri

> Youtube does state that explicit materials are allowed for educational content, but if their copyright machines can't even distinguish white noise from copyrighted music then I have very little hope that context like "educational discussion" will ever be recognised automatically. But why the obsession with recognizing it automatically? I've long wondered why we never developed a digital system for asserting copyright rights. Like, you register a piece of IP with Google and Google should give you a key. You can then use that key to sign certificates that grant certain people legal use of your content on Google. When I go upload a video, I upload all of the relevant keys that indicate that I've included certain copyright material in my video but I've cleared it up with rights holders ahead of time. Video game companies, for example, could one click hand out keys for free to make sure game streamers could legally stream their games. Similarly, Google could outsource their enforcement of "for educational discussion" to individual organizations. Cornell could agree that staff uploading content that is in keeping with Google's TOS is an official part of Cornell disciplinary proceedings. In return, Cornell has to place a bond at Google that would be forfeited if they ever broke their agreement. Insurance companies could insure such bonds by evaluating whether Cornell has the right processes in place that would prevent TOS violating content from ever appearing on Cornell's YT channel. In return, Google agrees to turn off all ML content moderation on Cornell's account. This could easily scale down to smaller channels. If I'm a video essayist channel, I could agree to take a mandatory training course and put down a $X00 bond to have my channel be free from all ML copyright checking up to my first X0,000 followers/X00,000,000 views. If a human judges me to have violated the TOS, the $X00 is there to pay for that human's time. No spam channel would find it worthwhile to go through verification and they would still be subjected to the normal ML checking but people who depend on YT for professional reasons could be more certain of avoiding automated false positives & negatives.