The real issue is their stubborn refusal to understand that there even *is* a gap to bridge, that they maybe aren't as excellent at communication as they believe themselves to be. I'm calming down on this whole situation, but this is a fundamental problem Bluesky Team can never fix on their own.

“Bluesky claims to want decentralization and composable moderation, but they still enjoy abusing the power of arbitrary banishment.” - Link

There is something so telling that 「 free speech absolutist* 」 techno-libertarians who had to be dragged kicking and screaming just to implement blocking nvm any sort of ( shitty ) moderation to protect people *IMMEDIATELY* turned into overt fascists exercising dictatorial power when they got it

None of these journalists will address the specifics of what happened in 2023 on here (citing the expertise of Sydette, Aveta, Link), and none of them will talk about the entire shaping of this platform around excluding Palestinian users. This is because of racism. Thanks for coming to my ted talk.

A great way to turn the temp down here as @jay.bsky.team @pfrazee.com and @aaron.bsky.team want would be to stop communicating site policies and practices with individual users or on their timelines and hire a comms person who is a professional communicator to do all the front facing interaction.

THEY APPARENTLY DIDN'T EVEN TWEAK IT. So... Outsourced without reviewing the product, taking at face value someone else's tech is all-knowing and infallible because they said so, job done, I deserve a medal for SSS-tier moderation! Bragging about not doing a job that they also want praise for.

Starting to believe you guys have an interview process where you make sure everyone you hire is absolutely utter shit at explaining anything about the site, so every response and skeet anyone posts just makes every issue worse. This is literally the dumbest response you could have given.

The thing with bluesky using AI to moderate images isn't that it's training on them, think for 10 fucking seconds about it, but that they are saying "yeah we outsourced that shit entirely, we got no control over it. It says label, we label" etc

tbf this is basically how all automatic content moderation works but I would have not stepped on the rake of saying "oh yeah we just completely outsource our decisions to this company's AI and don't review or tweak it at all"

They're eternally fuming over not attracting the genAI-gooner crowd, the "still buying NFTs 3 years after that shit died"-crowd. Those are the kinds of freaks that are not only with techbros shitting in their mouths, but actively grateful for the pleasure. Instead they got PEOPLE.

Oh GOD we're all just gonna get hit with the "anti-AI-racism"-label now for talking shit about Hive which is a real person and Aaron's friend uwu why won't we stop hurting their feelings 😢

now THIS is trust AND safety

So an anime girl with prosthetics was labeled "explicit sexual content", & then we learn that prosthetics are commonly labeled this way, & then we learn that automod has been outsourced to this ai company, & THEN we learn that the ai company might be using all our images to train ai & mods go silent

WHAT ARE YOU DOING MAN

Holy sweet fuck that's significantly worse. This also explains why Black women get hit with harsh labels for simply posting pictures.

Damn, can't believe Aaron's WE ARE FUCKING INCOMPETENT AND WE DON'T HAVE REAL JOBS AND WE TAKE NO RESPONSIBILITY ALL BLUESKY MANAGEMENT IS DONE BY AI WHICH WE ALSO DON'T REVIEW BRLBRLBRLBRLBRLBRL I'M A DROOLING MORON AND EAT MY OWN FECES excuse didn't calm people down and resolve the situation.

So you just leave everything to an AI while you, what, sit around and drink coffee and play games on your phone? What is the purpose of your division if there is no accountability?

They've banned several users critiquing them in the hopes the issue will be forgotten

So a computer is making management decisions and thus no one is accountable?

Hello again @aaron.bsky.team. Does Bluesky has a special contract with thehive.AI where thehive.AI won’t use Bluesky users posts for their GenAI? Or is Bluesky actually feeding every user post (Art & Writing) to a GenAI system that profits off of copyrighted material not intended for it?

aaron "gooning to barely legal on the clock" rodericks, everyone!

On a recent post you admitted mods only have 30 secs to review reports and now you have admitted to using AI on every image posted on this site that gives false positives, thus increasing the workload of these mods to correct these mistakes. It really shows you and your team are unprofessional.

Really shouldn't be a surprise to anybody who's had images auto-labelled incorrectly, but here's confirmation that yes, (and as I have stated in the past,) every image you upload gets AI'd.

"A critical part of our moderation infrastructure is a third-party black box that we don't provide input to" is, uh, kind of concerning.

Hey artists, how do you feel knowing your art is being fed into an AI slop generating machine so that disabled people automatically get labeled as porn on this website? Do you feel good about leaving Twitter just to have another company feed your art into the slop machine on purpose?

bro gets paid to be made fun of and do nothing

It's fun abdicating responsibility to AI, isn't it? And if it can train off people's work, even better. Your team is laughably awful.

HOW CAN YOU NOT MANAGE ONE SINGLE FUCKING DAY WITHOUT MAKING AN ALREADY VOLATILE SITUATION INFINITELY WORSE?! And now I gotta just ask: WHAT THE FUCK IS YOUR ACTUAL JOB?! Since you clearly just claimed to not be doing any "Trust & Safety"-work at all, the all-knowing & infallible AI does it! 🤢🤮

"WE DONT ACTUALLY PROVIDE ANY INPUT INTO OUR MODERATION" SAYS HEAD OF TRUST AND SAFETY

Bluesky isnt responsible for Blueskys moderation because the Bluesky vendor Bluesky hired to moderate Bluesky doesnt answer to their client, Bluesky Is one helluva justification

Bluesky head of trust and safety confirms they don’t provide any input into the moderation. Also confirms all of it is being fed into an AI… bro what the fuck.

Is that good

You don't understand dude, they have a huge backlog of posts to review, going back months. So if you post a joke to the CEO you get permabanned in hours, because of that huge backlog. That's the "trust" in trust and safety. It's how he's built so much credibility with the user base here.

So Links ban due to an image that had been up for 5 days, had no input from you guys? The ai was just out getting coffee for 96 hours?

An elderly German Jewish man was arrested for comparing Gaza to Auschwitz. His relatives were killed by the German state in Auschwitz. Talk to me about how Germany was denazified www.instagram.com/reel/DPL44U4...

The Bluesky Head of Trust & Safety admitting that all posted images are processed through an AI autolabeller that presumably employees have no access to or control over? (And probably steals every image to be used in an AI “art” generator, who knows?) Seriously, this is outrageously irresponsible.

So you guys have an automated system that falsely flags marginalized people and you have moderators that make decisions that can ruin people's livelihoods in 30 seconds. Sounds like your systems kind of suck.

What in the actual fuck is wrong with this place Jesus Christ

Absolutely bizarre just how blithely you'll admit to regularly auto-labeling disabled people as porn

If that were true I doubt link would have been banned for an image he posted 5 days earlier

they cannot let youtubers make books

Like, putting aside the fact that Moses Norton is a genuinely shitty human being, would *you* spend a whopping $48 USD (pre-tax and shipping) on a book that cites a bunch of fan wikis and claims it’s in-depth research? Would you buy another book from a pub who A-OK’d this? bsky.app/profile/taka...

But users and posts don’t matter, you’re not running a social media business, right? It’s all about the ATProtocol. Right??

Heute ging ja auch irgendein Statement von einer Wirtschaftslobbygruppe rum, dass die Chatkontrolle langfristig der Wirtschaft schaden wird. Als ob das die Blackrot jemals gestört hätte wenn einige Branchen den Bach runtergehen

How is this still up. This will be shown as an example of what not to do in PR for years to come.

Oh, wait. They already went back on that statement they released, saying that they'd come to an agreement with the author that 100% of his royalties would go to a charity? Well, any goodwill the publisher earned from that initial decision is gone now.

Es ist übrigens sehr unplausibel dass eine Regierung die plant Queers anlasslos auf Polizeilisten zu setzen davon zu überzeugen sein wird dass anlasslose Massenüberwachung mit der vorgeschobenen Begründung "Aber unsere Kinder!!!!" irgendwie nie schlechte Idee wär. Und damit meine ich die SPD.

Ich hab' legitim mehr Hoffnung dass die deutsche EU-Fraktion unter Druck einknickt und doch noch aus eigenem Willen gegen Chat-Kontrolle stimmt als dass die Bundesregierung auch nur einen einzigen feuchten Scheiß gibt heute. Und selbst die Hoffnung ist "nahezu null", aber es wäre zumindest MÖGLICH.

Bald könnten ALLE Chat-Nachrichten von den Behörden durchsucht werden – selbst ohne Verdacht! Das entscheiden *heute* die Minister*innen. Sag Nein zur Chatkontrolle! Jetzt unterzeichnen ✍️

This screenshot from Twitter (his username is Well-Dead Mage) did the rounds, and he apparently has a history of being like this. The deleted post I quoted is from the author's now deleted account on here. Oh, and I wasn't the first person to bring attention to this screenshot.