How often is this happening? I wonder if your site is being specifically targeted for some reason.
I’m taking a wild guess that it has something to do with one or both of these sites: http://bestpickleballpaddle.com/, https://thepickleballpaddles.com/. (Note, those are real sites, but let’s not give them too much traffic from here.)
The most obvious motivation I can think of for generating a post like the one in your screenshot is to gain some trust on the site in order to post spam links in the future. If that’s the issue, it might be worth changing the value of your site’s min trust to post links setting from the setting’s default value of trust level 0 to either trust level 1 or 2. If there are some domains that you’re happy to allow users of any trust level to post to, you could add those domains to the allowed link domains site setting.
I’m just spitballing here, but I’m wondering if instead of putting a lot of effort into trying to catch content generated by LLMs, it would be better to try to reduce the motivation for creating those types of posts in the first place. My concern is that LLM generated content will become more difficult to detect over time. Falsely identifying content as having been generated by an LLM could become an issue.
Another approach that might be useful for some cases would be to use membership in an organization, or activity on another platform as a criteria for creating a Discourse account, or as a criteria for gaining a trust level that allowed posting links on a Discourse site. For example, your signup form has an optional field asking for the user’s USAPA Rating Level. If membership in the USAPA gives members a profile page on their site, you could make membership in the USAPA a requirement for signing up for your site by having “USAPA Profile Link” as a required field on your signup form. You could then enable the Discourse must approve users site setting and only approve users after you had confirmed their USAPA membership. Another possibility would be that Discourse could add a feature that would allow a user’s trust level to be limited based on a custom criteria. For example, don’t allow a user to progress past trust level 1 until their USAPA membership has been confirmed.
I’m just using USAPA membership as an example here. I can see why you might not want to require USAPA membership for users on your site. The idea of establishing trust via membership or activity on another site is related to the plugin that’s being discussed here: $10k Bounty: Gitcoin Passport plug-in for Discourse.