#moderation — Public Fediverse posts

You know what *might* really help reduce spam sign-ups on Mastodon? Let instance admins pick their own security questions.

To register on oslo.town, you must answer: "Hvilket område ligger Barcode?"

To register on toot.wales answer: "Cwblhewch enw'r orsaf drenau ganlynol: Llanfairpwllgwyngyllgogerychwyrndrobwllllantysilio____"

To register on k-pop.social answer: "Name the missing BTS member: Jin, Suga, J-Hope, RM, Jimin, V, and ____"

#Mastodon #mastoAdmin #MastodonAdmin #Moderation

A quotation from The Bible

Wine is very life to humans
   if taken in moderation.
What is life to one who is without wine?
   It has been created to make people happy.
Wine drunk at the proper time and in moderation
   is rejoicing of heart and gladness of soul.
   
[ἔφισον ζωῆς οἶνος ἀνθρώποις,
   ἐὰν πίνῃς αὐτὸν ἐν μέτρῳ αὐτοῦ.
τίς ζωὴ ἐλασσουμένῳ οἴνῳ;
   καὶ αὐτὸς ἔκτισται εἰς εὐφροσύνην ἀνθρώποις.
ἀγαλλίαμα καρδίας καὶ εὐφροσύνη ψυχῆς
   οἶνος πινόμενος ἐν καιρῷ αὐτάρκης·]

The Bible (The Old Testament) (14th - 2nd C BC) Judeo-Christian sacred scripture [Tanakh, Hebrew Bible], incl. the Apocrypha (Deuterocanonicals)
Book 22b. Sirach (Ecclesiasticus) 31:27ff (Sir 31:27–28) [tr. NRSV (2021 ed.)]

More about (and translations of) this quote: wist.info/bible-ot/85500/

#quote #quotes #quotation #qotd #bible #oldtestament #apocrypha #sirach #ecclesiasticus #alcohol #drinking #enjoyment #happiness #intoxication #merriment #moderation #wine #party

#Moderators #Moderation

If there is a cyber attack, please let me
know about it, asap, since that is why i
chose to leave #BlueSky, after they got
attacked over the weekend, by "someone" (MAGA), who blamed it all on Putin's Iran.

#BlueSky has been compromised by
#RedDawn (extortionists), so if they
try that on here, with #Mastodon, then
i will go find some other social media.

Any attack would likely be aimed at me,
by lawless bad actors of hostile intent.
#LetMeKnowAndIWillGladlyLeave💝

In my own experience, a lot of people just treat server rules like terms of service. They don't read them, sign up, break the rules, get moderated, and then move on to the next server.

Some people even join servers that primarily use a different language and seem to expect that they won't be moderated there.

Sooner or later, though, they usually get kicked out unless the server is poorly moderated.

#selfHosting #moderation #mastoadmin #fediverse

#MissKittyRaw
I love getting added to #AI #moderation lists when the #documentation I deliver is as good as you can find on the network. I just love people that are unwilling to learn, and think that they know something, and they live inside of a delirium created by non-engagement with #facts.

Discord AI moderation bug wrongfully bans 8,000 users over spreadsheets and chessboards

Share if you found this interesting.

https://1ban.news/discord-ai-moderation-bug-2026/

#1ban #discord #moderation #bug #tech #technology

Alegar propriedade dos alvos ou participação em testes de segurança costuma bastar para contornar barreiras de modelos de IA, segundo pesquisadores da Cisco Talos, que analisaram abusos por cibercriminosos. (EN)

https://www.theregister.com/security/2026/08/04/bypassing-ai-guardrails-is-so-easy-a-script-kiddie-can-do-it/5282973

#ai #cybercrime #cybersecurity #moderation

RE: https://oldfriends.live/@paul/117068557864169414

Some anal person reported this as offensive because of the TV character's name, Crazy Cooter. I guess they don't know that cooter is a term in many states, esp the south, for freshwater turtles. Some people need to get over themselves and their destructive need to find fault or offence in every feck'n thing and police it. #Moderation

Cincinnati Zoo: Common cooter: https://cincinnatizoo.org/animals-archive/215750/

Massachusetts: Red Bellied Cooter: https://www.mass.gov/info-details/northern-red-bellied-cooter

Florida wildlife and game: https://myfwc.com/wildlifehabitats/profiles/reptiles/freshwater-turtles/cooters/

@ashwin #Answers:

3/ I think #lineage can reduce possessiveness and acquisitiveness - to a similar degree that we have it stamped and forced in today.

But it needs a similar kind of #love "#force" which might in the opposite #spectrum of the #extreme (where you "must" share) while being softer in "#enforcement" (#guidance / #moderation / #testing).

And perhaps always being customisable by #individuals after also, according to where they are and their #different #selves / personalities.

Wie kommt man zu einer eigenen #Instanz?

Das #Fediverse besteht bekanntermaßen aus vielen, vielen einzelnen Instanzen und basiert auf offenen Standards. Daher besteht die Option, sich eine eigene Instanz aufzubauen und mit dieser am freien Social Media teilzunehmen.

Nehmen wir mal an, Du besitzt nicht zufällig nebenher ein ans #Internet angeschlossenes Rechenzentrum und mietest entsprechende Infrastruktur bei einem #Hosting-Unternehmen an. Ob nun "Shared Hosting", einen virtuellen oder physikalischen #Server... wir gehen hier mal davon aus, daß Hardware, Betriebssystem(umgebung) und Netzwerkanbindung nicht selbst gelöst werden muß.

Dann fallen spontan die folgenden drei Möglichkeiten auf den Tisch.

1) Da #ActivityPub ein offener Standard ist, besteht die Möglichkeit, sich die entsprechende fediversale #Serversoftware selbst zu entwickeln. Das ist ein klein wenig ein Hardcore-Ansatz und aus eigener Erfahrung kann ich sagen, es dauert... ;-)

2) Da freundlicherweise viele fediversale Serversoftware als "Open Source" entwickelt wird, schaut man sich nach einem geeigneten vorhandenen System um und installiert und betreibt dies anschließend selbst. Das kann dann eine Instanz vom Typ #Mastodon sein oder #Friendica oder #Firefish oder #Pixelfed oder...

3) Man schaut sich nach einem spezialisierten Anbieter um, der auch die Installation einer fediversalen Open-Source-Serversoftware für einen erledigt. Die existieren... bieten aber nicht zwangsläufig sämtliche existierende #Software an.

Allgemein muß man sich noch entscheiden, ob man die eigene Instanz nur für sich selbst nutzen will oder Konten an Dritte geben will.

In jedem Fall muß man sich darüber klar sein, daß der Betrieb einer öffentlichen Plattform im weltweiten Internet ein klein wenig Punkte zu bedenken gibt. Sei es zum Beispiel juristische Auflagen (z.B. Pflichtangaben im #Impressum, eine Datenschutzerklärung, die Einhaltung der #DSGVO usw.). Dies gilt insbesondere, wenn Dritte die Plattform nutzen dürfen sollen. Dann kommt auch noch der Punkt der #Moderation hinzu.

Alles in allem, nichts für eine spontane Fünf-Minuten-Idee, aber auch nichts, was völlig unmöglich erscheint. Etwas (längerfristige) Motivation, Zeit und Geld... dann gilt das Motto eines schwedischen Möbelhauses: Entdecke die Möglichkeiten! :-)

So every once in a while, I open the Administration panel on my Mastodon instance and find a huge pile of reports waiting for me. 😅

Sometimes I also get messages like, “Please moderate signups for the time being,” especially when spam campaigns are actively targeting instances with open registration.

And I’m actually working on it!

I’ve now switched Mustard back to approval-based registration and introduced a new signup review system. The main reason I hesitated to require approval before was that I couldn’t always respond quickly, which sometimes left legitimate users waiting too long.

The new system sends signup requests to me through Matrix, so I can review and approve them much more quickly without constantly checking the Administration panel.

It’s still a work in progress, but I’m hoping this will make Mustard both safer from abusive signups and easier for legitimate users to join.

More improvements are coming. Thank you.👍

#Mastodon #Fediverse #SelfHosting #Moderation

In Ars Technica:

❮… “And there was nothing we or the experts [who posted the deleted content] could do about it,” Dr. Sarah Gilbert, one of the mods, told me. …❯

As a co-moderator of a different subreddit, I must say, that seemed unlikely.

There'll be knee-jerk reactions (in Ars Technica comments and elsewhere) from haters of Reddit and haters of AI; I'd like to better understand the truth.

Postscript: <https://mastodon.bsd.cafe/@grahamperrin/117060193842963024> confirms that removals occur for new comments that link to the rarehistoricalphotos.com domain.

Numerous older comments do include links to the domain.

#AI #moderation #Reddit

Cc @patrickcmiller @arstechnica

How many accounts say things like #NoHate and then #block me and add me to #moderation lists? LOL.
How come they didn't ask me why I haven't built my own #decentralized #genAI which wouldn't have #climate impacts? Why didn't they ask me why I use it? Cuz people think they know something w/o work.

This is great!

"The way we handle reports is proportional, context-aware, and, wherever we can, aimed at helping people put things right rather than simply penalising them."

https://help.joinmastodon.org/article/20-moderation-decisions-and-appeals

And also, an official guide on how to make collections on Mastodon!

https://help.joinmastodon.org/article/19-creating-a-collection

via https://mastodon.social/@staff/117054818385789452

#mastodon #fediverse #moderation #transparency #SocialMedia

Disposable and custom-domain email addresses can be really tricky to deal with as an admin of Mustard, because sometimes you simply can't tell them apart.

We've had plenty of spam accounts that randomly mention users, as well as AI bot accounts that appear to be scraping data from the server. At first, you can usually spot the suspicious ones pretty easily. But after a while, you start seeing accounts using all kinds of unusual email domains, and it becomes much harder to tell who's legitimate.

Sometimes I've ended up suspending accounts simply because their email domains looked suspicious—even when their profiles, including their profile pictures, looked like they belonged to real people.

It's one of those #moderation problems where you don't really have enough information to make a confident decision.

#selfhosting #community #mastodon #mastoadmin

What PieFed’s “social credit system” actually does

Just like Reddit, PieFed keeps track of how many upvotes and downvotes your posts receive and the net value of all the votes received is called your “reputation” (Reddit calls it “karma”). This is not shown in the UI anywhere (except to administrators), to dissuade the karma-farming behavior common on Reddit. Reputation is not a game or status symbol, it is a mechanic for enhancing trust and safety on PieFed.

Some people have criticized PieFed for this, calling it a “social credit system” similar to the one in China so I thought it’d be good to explain exactly what it does so people don’t need to guess or rely on hearsay. 

Background

Recently I did some analysis of voting patterns and found that people using PieFed upvote posts 33 times more than they downvote. We’re a friendly bunch, we want to encourage good contributions, we upvote all the time. In this environment it should be very very easy to have a reputation above zero.

Despite this supportive environment, some people (around 1%) have a negative reputation balance.

People who get downvoted a lot tend to be posting spam, trolling or being exceedingly negative. Sometimes a well-meaning person with strong and well-supported opinions will go ‘against the herd’ and receive unfair downvotes but if they stick around long enough the upvotes they receive for other activity will easily balance this out.

How reputation is used

Spammers

PieFed uses reputation to spot spammers, producing a report to administrators showing the most downvoted content in the last 3 days. This helps admins take action before people report the posts. Consequently, piefed.social is one of the least spammy places on the fediverse.

Administrators can also sort the user list in the admin area by reputation, to find problematic accounts. Also when viewing a list of new accounts, any account with a negative reputation is worth taking a closer look at because it’s usually a spammer. 

Trolls

People who get downvoted a lot have a warning icon next to their name, like this:

This warns others that they might not be worth engaging with. The icons are not shown to the person who earned them, to avoid inflaming the situation.

At -3 reputation it’s one orange triangle, at -12 it’s two red triangles.

Roughly 0.4% of active accounts have two red triangles on them. It is exceedingly rare. Most people like that crash around making a mess everywhere then lose interest and go away or get banned.

Direct / private messages

At -10 reputation or lower the account can’t initiate direct messages (although if someone else starts a conversation with them they can reply). Abusive people, who tend to have a low reputation score, often use direct messages as a tool for harassment so this is a safety mechanism.

Trustworthiness

This builds on reputation a bit.

An account is ‘trustworthy’ if they have +100 reputation or are older than 7 days old.

Only trustworthy accounts can initiate direct message conversations. This stops spammers from sending direct messages and stops abusers from making a burner account to send abuse from.

The bio of untrustworthy accounts is not shown to anonymous viewers, to disincentivize spamming links there.

The way PieFed uses the concept of trustworthiness is quite limited. I see this as the beginning of something like what Discourse does but it is not a priority area of development at this time.

Low quality communities

In some communities, which the instance admin designates as ‘low quality’, upvotes on posts do not increase reputation (upvotes on comments on those posts still do, though). Someone who exclusively posts in low quality communities might have a hard time maintaining a positive reputation as downvotes decrease it while upvotes do nothing.

This feature has not turned out to be very useful (I just ban / delete really bad communities) and it’s not used much in practice.

And that’s it

Ultimately, reputation isn’t some gatekeeping tool for participation. It’s just a simple mechanic to help us filter out spam and harassment, while keeping moderation decisions in the hands of real people. To 99% of us, it’s completely invisible. 

In the past there were a few more niche situations which dinged accounts for 1 reputation point here and there but it was ineffective and inconsequential (compared to the flood of upvotes most people receive) so those were removed.

I hope this clears things up for people. If you have any questions, let me know in the comments! 🙂

Tired of Discord turning into a propaganda sewer? Meet Miderator - an open-source Mistral AI mod bot 🇪🇺.

Auto-filters:
🇪🇺 Anti-EU rhetoric
🇺🇸 US/MAGA
🇷🇺 Russian/Pro-War
🛢️ Anti-Ecology
🏳️‍🌈 LGBTQ+ hate
🧳 Immigrant hate
♀️ Women hate
💉 Anti-Vax
📿 Religious Propaganda

Open-source = fork, contribute, defend your space. Solidarity through code 🛠️.

Add to server:
https://discord.com/oauth2/authorize?client_id=1534482774054932723&permissions=76800&integration_type=0&scope=bot+applications.commands

Code: https://github.com/ArtemkaKun/miderator

#Discord #OpenSource #MistralAI #EU #Moderation

Mistral's Shieldstral: 3B open-weights model for multimodal moderation

https://mistral.ai/news/shieldstral/

Comments: https://news.ycombinator.com/item?id=49171268

#HackerNews #Mistral #Shieldstral #multimodal #AI #model #open #weights #moderation #machine #learning

Wenn MissChriss auf die Bühne geht, muss vorher so einiges geprüft werden.

Und am Samstag wird auf dem Jungfernstieg geprüft - kommt alle vorbei!

#pruef #demo #drag #soundcheck #moderation

Reminder that you can, and should, mark anything talking in detail about or depicting disturbing subjects as "Sensitive" AKA with CW, using the built in features. This especially includes #war, #genocide, snd other #violencr and #bigotry. Just click the button at the bottom of the post window and add a brief and vague description of what you're talking about that is upsetting. At least give us a moment of warning.

Not using CWs is a quick way to get blocked or limited by a lot of users and servers, and generally speaking, people have enough to deal with. Give us all a break will ya?

- TechHub #Moderation

#FediTips

9.0.0 — Growing Up https://activitypub.blog/2026/06/12/9-0-0-growing-up/ #ActivityPub, #BlurHash, #Fediverse, #Moderation, #StarterKit, #StarterPack, #WordPress

RE: https://mastodon.green/@madiko/117015488276016186

Insbesondere grüße ich @JohanEmpa, Admin von mastodon.green. Es ist schön, dass es Dich gibt und Du unser System hier am Laufen hältst. 🥰

#SysAdminDay #SystemAdministration #Moderation #Fediverse

Allowing the recovery of users accidentally deleted by moderators - A new issue for @pixelfed

It would be helpful if a user deletion could be undone, as is the case with Mastodon or Lemmy. It would be sufficient for a user to be restored within a certain period (30 days?).

An alternative could be to add an intermediate stage between the active and deleted accounts, namely the blocked (or frozen) account. This way, the user and all their content are made invisible until the administrator can get feedback from the user.
This way, even if the user were accidentally blocked, the operation would be reversible.

Pixelfed is enjoying great success, and we administrators are doing everything we can to contribute to this success, both through careful and ongoing moderation and by keeping registrations open. Unfortunately, this has led to an influx of spammers, and one way to address this is to mass-delete all suspicious users based on their username, language, and other external factors.
The inability to restore accidentally deleted users, however, limits this solution and forces us to spend a considerable amount of time manually checking each suspicious account.

https://github.com/pixelfed/pixelfed/issues/6662

@fediverse

#Pixelfed #Moderation #Fediverse

Cuz nobody ever predicted this right? *Cough cough* #Skynet *cough cough*

Second major #AI company says its systems hacked into other firms

#Anthropic disclosed the previously undetected incident one week after #ChatGPT developer #OpenAI said one of its systems hacked a #tech company.

#GiftArticle
#law #regulation #oversight #moderation #HumanElement
https://wapo.st/4g38Gml

What 20 Million Bans Reveal About The Stress On Wikipedia’s Volunteers

https://fed.brid.gy/r/https://www.techdirt.com/2026/07/30/what-20-million-bans-reveal-about-the-stress-on-wikipedias-volunteers/