A lot of sites based in Europe are going to start requiring systems like this with the advent of new safety laws, so I'm always happy to see attempts, but there

A lot of sites based in Europe are going to start requiring systems like this with the advent of new safety laws, so I'm always happy to see attempts, but there's something I can't quite get my head around. We've recently seen social media sites get attacked by people posting abuse material, as in this news story: https://www.bbc.co.uk/news/articles/c93qw3lw4kvo https://www.bbc.co.uk/news/articles/c93qw3lw4kvo .. how do/can NSFW detector services help with this problem? Filtering out bikini or porn seems easy enough. But what if someone starts uploading the most offensive of material? Is the model trained on that? If so, how, legally? Or if using an external provider/API - are they going to be reporting you to the authorities when such material gets sent to them for moderation? It seems like a legal minefield, despite being a sorely needed service.