We’re definitely getting some AI bots around here these days. We don’t really get any new users so when we do it’s pretty easy to tell that they’re bots from their first post - always some bullshit non-actuarial comment.
But it’s going to get worse, to the point of being overwhelming. Right now it’s fine, but eventually it won’t be.
I think there’s sort of two ways to slow down the bots. One is to manually approve everyone instead of auto-registering. This isn’t my preferred route because it’s more work and we don’t really know that they’re bots until they post anyway.
Secondly, we require a corporate or non-gmail email account. That solves the problem, but is more problematic for new users. I’m leaning this way mostly because it’s ‘what new users?’.
Could you do both – require a non-free email account OR manually approve if using a free email account?
Or, as an alternative…would it be possible to switch to a “send us an email with a subject containing the word [ x ] to register” model, with a bot autoresponding with a registration link?
That’s work for me on an ongoing basis plus requires that we modify the discourse software, drifting us away from the free version. Both are nonstarters.
i’m not doing anything right now, but I think soon we’ll have to move on this. Mostly just looking for deal breaking concerns about requiring a real email account. Anyone interested in the site will have one.
Well I guess in the end it doesn’t matter, because the number of new registrations is pretty much zere anyway. when was the last time we had a new poster? 2025? Everyone who’s coming on board is already here I think.
There’s the ability to have the “Trust Level” set to 1 which requires a moderator to approve any post the user makes.
That can allow any moderator to review–and then set the Trust Level to 2 if we feel confident that it’s not a Bot–and not really add more work to your end.
