"should use AI . ."
Indeed. If only we had AI, there would be a great many things we could try and apply it to.
What I find interesting is the "social sites should police themselves" attitude. Neatly allows you to wash your hands of the issue, doesn't it ?
You don't ask supermarkets to "police" their customers, do you ? Malls do not have SWAT teams. So why should it be up to social sites to decide what to take down ? Youtube does takedowns based on DCMA requests and that is generally very badly handled because there is very little recourse. You want to see that reproduced on FaceBook and Twitter ? I don't use them and don't really care, but I am convinced that it is a governments' job to force the application of the law.
So someone is not happy about something that has been published, he goes to the cops and files a complaint. The cops check that the illegal nature of the thing is true, then send a takedown request - which is mandatory because The Law - and the site complies, alerting the original poster that a specific piece of his published data has been placed under legal lockdown. If that one is not happy about it, he can file a counter-claim and go to court.
Of course, depending on the level of illegality, he might not even need to go and file a counter-claim ; the police could be knocking at his door to tell him in person that what he posted was not a good idea and could you please follow us, sir ?