I think the only thing I've seen universally agreed upon to be bad is child pornography.
Compare and contrast with how this thread, instead of focusing on "lol, algorithms and context!" has turned into questioning whether the real problem is advertisers signalling they don't want to attach their brand to racist videos (and let's not get started on how angry much of HN was when companies decided to manually enforce policies against hate speech). It's difficult to conclude that most US "free speech" advocates don't believe that banning [perceived] racial hatred is somehow more dangerous to humanity than banning [perceived] adult content (which parallels Supreme Court rulings on what speech does and doesn't deserve protection). Given the respective consequences of racial hatred and pornography, I see this simultaneous presumption in favour of the former and against the latter as very hard to justify.
For example, I'm told the character creator in 'Cyberpunk 2077' offers adjustable penis size - but many streamers and reviewers won't risk showing it. Even comedy channels that love creating zany characters.
After all, would you trust Youtube to detect whether nudity is pornographic context, when they can't even tell if 'white threatens black' is in a chess context?