Good question.

Right now we run some ML models to check for illegal content and then respond immediately with the ban hammer.

We also monitor content published on our platform with some admin tools we built.