Hey peeps, the well-known spam problem in lemmy DMs is getting slightly worse, and the spammer in question is evolving their tactics to evade the very rudimentary options we have built-in to lemmy itself (or should I just call it,non-existent?). To get ahead of this, we deployed a DM-scanner directly in the DB, based off of this code but adjusted so that it's more difficult to pull off shenanigans.
At the moment we're only deleting messages based on the "fediverse chick" spam. However I want to point out that if for some reason you legitimately DM someone on dbzer0 with similar terms, you DM might get deleted, so do be aware about that. Them's the breaks.
Unfortunately I can't reveal the exact code I'm using atm, as this can lead to the spammer adjusting their tactics to evade it. However I plan to adjust threativore to also be able to manage your DB anti-DM-spam filters in the future, to make it much easier to handle this, even if you don't feel confident touching your DB. Stay tuned.
I do hope we get more robust anti-spam tech in lemmy and I did open two issues about this, but it seems this is not a prio atm.
If you realize that the anti-spam ate a DM. Do let us know. If you see spam getting through, also let us know.
