I would prefer something like moderator queue, but on top of it, we can also use help of AI to mini moderate the queue, so if ChatGPT thinks that post are more of a spam thread, it moves it to the bottom of the moderation queue (to the back of the line), so it takes way longer for an mod to get there, while genuine post stay at it's place, so those get approved quicker. In the case GPT didn't catch the spam thread, a moderator would have removed it anyways. The Spam thread would get moderated anyways, but the mods would have fallen asleep to even consider approving it. (Don't worry the queue is sorted by time, so everything would get moderated, (yesterday's queue would have priority).Also, just to make spammers life harder, removed threads/bans are not instantly shown, until 72 hours later because just why not. Also we could just make it instead of you're banned, it just says "your credentials are incorrect" or "please complete the captcha". (and of course any removed thread is also sent to GPT as feedback and improvement)
This, if implemented, spammers would have little incentive to try to bypass GPT because the post is still getting moderated. If they did, it doesn't improve their chances of getting the thread posted or kept there. Regardless of how many aged accounts or new accounts they created, their post would need to pass an actual moderator, not an automated system.
I am not sure if we do default show or default hide: default show means all posts are shown, but is waiting in the queue to be moderated. default hide means new posts are not shown, until they are moderated in the queue.