only one data point, but fwiw when I worked for Google I found some actively toxic youtube content w/upwards of 500k views that was telling children to off themselves, and despite using my employee back-channel connections the most I was able to get was an eventual "I'm not allowed to do anything about this" from a YouTube moderator, though it seemed to be for technical reasons (all the nasty content was in annotations, which apparently weren't wired into the moderation pipeline). There definitely wasn't a red button for me to hit as an employee to get it taken down.