Lots of good suggestions here, let me throw one more in the pot -- could you do an equivalent of a "ghost ban"?
Instead of blocking their IPs, detect if the traffic is coming from the abuser's IPs, and serve different content -- blank, irrelevant, offensive, copyright violations, etc.
Instead of blocking their IPs, detect if the traffic is coming from the abuser's IPs, and serve different content -- blank, irrelevant, offensive, copyright violations, etc.