Youtube already has quite an interesting clickbait detection mechanism. If the click ratio on a thumbnail is high, but the watch time low, it will be downranked as clickbait.

https://www.socialvideoplaza.com/en/articles/what-is-clickba...

This means if the thumb/title is a bit sensationalist, but the content is still good, it could stay up. This leads to the current situation where slightly annoying titles and thumbnails are used, but the really egrious stuff gets filtered.