Yes, I understand that a regex isn't a long-term solution, but if you've actually seen any of these pop up on YouTube (they're on the front page of my recommendations right now; I've seen probably 30 just today), you'll know what I'm talking about. These are what I'd call "low-hanging fruit" spam – extremely easily distinguished from real content. After months of it happening (and I've seen literally hundreds of the exact same type of video recommended to me), a simple solution would undoubtedly be effective, or at least better than nothing. Get a single intern to double-check the ones that your algorithm identifies to avoid FPs. If the scammers get more sophisticated, cross that bridge when you get to it. But solve the immediate, easy issues now. It's seriously not that hard. This isn't the kind of difficult ML-generated spam that you're probably thinking of, and I agree - at that point it does become a game of cat and mouse. At that point, I'd be understanding when it sneaks through the cracks.

I've dealt with similar issues in my own work (I run a platform with a lot of content posted daily and have multiple tiers of spam control in place). The spam I'm seeing on YouTube is not the difficult kind, yet they're doing nothing to deter it. That's my complaint.

I've probably reported hundreds of these, so at the very least, stop recommending videos with the exact same title, description and content to me. It's ridiculous.