@iftas@mastodon.iftas.org seems like it's checked for @nodebb@fosstodon.org :shrug:

@iftas@mastodon.iftas.org in NodeBB, after clicking "reject" a million times (or in Mastodon's case, a checkbox, I guess), we added a "Reject All" button.

Does Mastodon not have that?

Now for spam review, we go through the queue and manually accept the ham, and one-click blackhole the rest.

@julianPelago is that specific instance of FIRES. I've updated the post to make that clearer, thanks for the note!@thisismissem

@iftas@mastodon.iftas.org is Pelago FIRES different from @thisismissem@hachyderm.io's FIRES?

@iftas ufff finally I was sad to see that server fall into oblivion

@iftas I'll believe it when I see it. They're buying out the regulators again.

@JohnnyEnzymeI wasn't sure if it was our list formats that were the issue.

Blocking domains is something I can do at the general user level, but seemingly I can't do it from the community I run. So in the past, I've had to block malicious bots by hand when they bother us. Fortunately that's happened only... once so far, I want to say.

Not sure what you mean by different file format.

@JohnnyEnzyme Do you need a different file format, or blocking domains is not an available option for you?

Unfortunately, it seems that I can't add risqué domains to my communities block list via current PF software. Perhaps in future?

@julian Thanks, great info, much appreciated! I don't think there are any right and wrong ansesrs here, just very interested in what everybody's doing.

FWIW there is no policy at current on activitypub.space that explicitly prohibits or allows LLM generated content.

So it's case by case and usually the volume of generated content means it ends up getting banned if reported.

Such is already the case for noisy non-AI bots on sites I admin. They get blocked only to decrease my workload lol.

@jdp23@neuromatch.social

@julian Yeah, I don't know what the right answer is here, which is why I asked it as a question ...it's tricky because it's fairly-widely adopted but not standardized and here isn't a lot of stuff encouraging bot developers to pay attention to it. It's certainly signalling intent, but it's a signal that's easy not to know about.

@jdp23@neuromatch.social while I applaud the intent with #nobot, that isn't a hashtag that is standardized anywhere as far as I am aware.

At most, it's signalling intent.

@oli perfect!

@iftas I'm just keeping these as advisories for now, anyway, since they are suspected, no recommendations yet.

@oli just so you know, these are suspected, not confirmed, we have a false positive tracker and of the latest 682 accounts reported we found 4 to be false positives, this is not a perfect science.

@iftas Updated. 790 accounts now, it looks like.

@iftas I pulled from here:https://about.iftas.org/library/suspected-portal-kombat-accounts/But I can add accounts from that spreadsheet that might be missing. It should de-dupe.

@oli are you pulling from docs.google.com/spreadsheets/d ?

@iftas Also grouping them here.https://pelago.1sland.social/report/evidence/kind/group/entity/baaf1741-4f64-410f-84b1-0c9e2f9366f4

@iftas I just pinged them on x, they were active an hour ago:x.com/wakest/status/2045128445

@iftas I just had a look there. There are a number of known scammer accounts set up on there, like Lillian, Tara, Elysia, and multiple versions of themselves.I limited zirk.us for now. I didn't block it because there are a number of follow relationships here that I don't want to break.Anyone on this server (hear-me.social) who gets an unexpected follow request from zirk.us ought probably not accept it.

@iftas My understanding is that med-mastodon.com is also permanently shuttered.

@oli @julian yeah, so I'm interested in what filters applications need, and a good future feature will be "filtersets" which don't affect the public API but do change the admin experience (e.g., you could create a "silence" filter that includes these named filters, so you can apply them all at once repeatedly, but to a consumer it's still clear you're recommending they take these specific actionsAnd remember: the datasets only contain recommendations, they're not prescriptive. Any consumer can decide to ignore your recommendations as a dataset producer and do something different.