Any solution to this problem will be a burden to someone but what about
- implement a system that flags suspicious activity(X amount of accounts being registered in the span of Y amount of time is suspicious). You have to have some metrics in place in order to make a relatively accurate guess on what's normal and what's suspicious. How many new accounts are registered daily? On weekdays? On weekends? On happenings? On holidays etc..
- flagged accounts are not allowed to post for a month, only lurk
- monitor lurking behavior - what threads they lurk, for how long they stay active etc
- if the account stays idle or displays any abnormal behavior - ban
- after a month they're allowed to make 1-2 posts per day in the top 2 threads they were lurking, with moderator approval
- every new account that didn't get flagged initially has a 1 week waiting period before they're only allowed 1-2 posts per day with moderator approval. If the account stays dormant and waits out the week - 1 more week. Lurk. If you register and start posting immediately you're probably not a good fit anyway.
Include a big apologetic warning during the registration process:
"Sorry for the inconvenience, but The Gay Pedophile at the Gates has forced us to tighten up how new accounts are allowed to interact with the forum"
With a link to this thread where new users can get acquainted with the situation at hand.
It's a bruteforce approach but it will make the faggot dedicate 10x the time before he can start spamming, Registering 1k accounts in a bunch and spamming everything is one thing. Timing registrations to avoid flagging, then waiting, then possibly waiting again, then only 1-2 posts/day...that's a whole other deal.
Also put a system in place that scans their posts for ai-generated content. AI slop is recognizable and there are good small LLMs that clock the vast majority of AI content with high accuracy. If he plans on "growing" the accounts that'd add one more layer - he'd have to manually edit the initial posts he has loaded to make them seem less AI slop or even worse - write them himself. For 2-3-5-10 accounts that's whatever. For 100 that's lunacy.
It would require tons of management on his part to get to the point where he can spam again. I've got other ideas for spam prevention but I'm sure you've thought of them already so I'll leave it at that.