A user, WaitroseCarpark, reports experiencing a mass spam attack on their forum, with around 150 threads created in 10 hours. Despite tightening spam filters, they had to temporarily disable user registration. NateDhaliwal and nat suggest using the Discourse AI plugin for spam detection, which WaitroseCarpark decides to implement. Canapin shares their experience with a similar attack, using an unofficial plugin and a rails script to delete recent accounts. pfaffman provides a link to a method for retroactively reviewing and deleting spam posts. opcourdis suggests checking the ‘Require topics and replies to be reviewed by staff before being publicly visible’ setting.
First it started with promoting airline helpline numbers, now it’s moved onto quickbooks help. Around 150 threads started in the last 10 hours.
I’ve found the discourse spam filtering pretty efficient and don’t usually have to delete any more than a couple threads/users at a time for it to stop but this one’s different.
They look a little like this (I’m not pasting actual content):
We do get customers suffering from spam attacks at various times, not just isolated to this week or next. It could be 1-2 communities this week, and then 2-3 communities next week, and sometimes none in a week. FWIW, we have new spammers every day on meta.
That said, I highly recommend using the discourse-ai plugin’s spam detection as Nate quoted above. You can use a cheap LLM like Gemini Flash and I can attest it works very well.
In addition to the other solutions : As your first reflex was to stop users from registrating I guess this spam came from very recent users and that this was not setup :
unofficial)