Well the mods cant do much if they're banned...

true

True xd, I've seen people spamming questionable messages and images and their account is still not banned, at least they got banned from the servers I'm in

exactly... arent pedos known for using links instead? since straight up csem would be too noticable? like how twitter once had that huge disgusting mega links problem for several years till banning that whole tag.

If all children are banned, all children are safe from the pedos

Reminds of the guy banned for posting a bonsai that looked like a butt.

@@whome9842 buns-ai

People who legit defend using AI moderation, probably would say that KZfaq Has shitty moderation in spite of it being bots for like 10 years

It's probably a human, but imagine looking at an image for 2 seconds before deciding yay or nay.

The "human" in question is just another bot,which they have named Human™️

Its scanned locally. Didnt you watch his last video?

From my understanding it's only pictures you share on discord not your PC. They have it in their Terms and Conditions agreement and an EULA.

@@kevinkaric1511 Because that's so much better

@@SubtotalStar850-uh8pg Why would you care about such thing? The only disadvantage or security regard you will have is when you're actually exploiting children. You being mad makes you confessing as a pedo.

its their platform, you know it's unencrypted and they have access to anything. if ur using discord for sensetive shit then ur retarded. i dont expect privacy when i wank my shit in a starbucks

Saying you're 12 or below triggers the AI to disable your account

@@MaahirMomtaz12 Someone needs to report the message, but yes.

Do you think people would do that? Lying in their internet about their age? Man, what a country

i like to remember how discord didn't solve the furry pedo issue

​@@Fernybun you're now finding this out?

That’s discord for you!!! Here’s the thing tho. If you are false flagged and discord mods think your content is CSAM (even if it ain’t) wouldn’t you just have law enforcement on your door step for something you didn’t even do? Or are law enforcement actually kinda lazy like NTTS said when it comes to this sorta stuff

@@user-ti2cu7um3b Just don't send stuff that look like CSAM??? How often is that a question you have to ask before sending? If you have to ask that question often, I think you have a problem.

data was from 2021

@@lilcrabbyboi3075 nahh if you look at a transparency report from E safety Australia. You’ll see they talk about discords CLIP model and it has 80% detection in servers, DMs are 70% and private servers are 40%

@@user-ti2cu7um3b Oh discord doesn't even bother to report you to the police, it's why the FBI and CIA have hundreds of assets on discord

im already banned :)

@@thefiskar oh oh there's a no no fapper

my account got disabled for a stupid reason and discord hasen't responded to me :)

I just got banned

@@cacamonkey445They’re most definitely flooded with messages from many, many other people too. I hope it all gets fixed fast.

Having to "censor" literal wood with phallic veggies is pretty hilarious, but it's absurd that discord is even at that point in the first place.

discord ai when you show them a log texture : NOOO HOW COULD YOU DO THAT discord ai when you send the same thing but with an eggplant on it : understandable , have a good day.

Their filter seems incredibly sensitive to large concentrations of skin-like colors. It probably thought that the tree bark is an undressed black person

I remember something similar. A friend wasn't able to send a picture of his glasses, so we'd add a picture of Toby McGuire to make it work.

One thing i still find funny the fact that a random bot for moderation does a better job then any moderation discord has

I feel like features like this should also require permission from the server owner regardless of if they have administrator perms. Don’t get me wrong, administrator perms are supposed to give you access to everything in the server, but there are some things that just should be optional like mass-banning. Your idea is good as well though

the owner:

no idea why discord can't use their brains.. not to joke on the video, but having this much power under one person is ridiculous. It's in the same exact vein as having some stranger/AI ban me for having conversations with my friends without reason or appeal. This stuff could never happen in person, so shouldn't happen on a social media platform meant for games.

yup and all 3 need to do 2FA

@@_.luminosity._ Leaving everything to the owner leaves the "hacked account" issue still active, if just harder to do since you need to target a specific person.

Are you naked in your selfies? What are you sending to your "friends" lil bro 💀💀

Im pretty sure it has to do with CSAM not selfies. You’ll be okay as long as you aren’t naked.

As long as there is no nudity you will be just fine

if there is, add a watermark and eg. drivers license​@@kamabokogonpachiro5038

there was not any nudity in that image shown in the video. its highly context dependant but i dont think you should show ANY skin if you want discord to not notice

and remove any and all no appeals rubbish forever that is the big bite in the ass at the heart of all of these systems they need a way to let you Prove their bot/bleary eyed sleep deprived minimum wage intern that just wanted to get to lunch wrong such as posting the actress's bio profile to prove their age as an example

I agree, only verified cases should be reported, but non-verified cases should not be, false reports are a serious crime and sometimes as bad as those pictures in legal terms.

It's also incredibly trivial to get around by just putting everything in a password protected zip archive. Sure, it wouldn't be too hard to open up and scan the contents of the zip archive if the password is provided alongside it, but that would require a far more sophisticated automated system for checking uploaded files. If an automated system to detect and ban offending users is put in place, it needs to be transparent in how it functions (such as the case with PhotoDNA), and there needs to be at least some form of appeals process for potential false positives. Neither of those seem to apply to this machine learning solution. By their nature, ML algorithms are completely opaque, and nobody can really know exactly how they arrive at their outcomes, and from the sounds of things, perma-bans are handed out before a human can even verify whether or not the flagging was legitimate, with no way to appeal said ban.

they will never reverse their course because this narrative has been used to push invasive technology and legislation since the industrial revolution. "protect the kids" will always be an effective tool for doing things that have nothing to do with protecting kids.

So far discords a 3.3 rating on the app store and all the top reviews are 1 star lol

Agreed. Then if it flags said photo a human should always have to review it.

@@fadeddragon123 Yeah, the way this seems like it'll work it's just presuming everyone as guilty and will lead to a ton of false positives, especially since a lot of characters look younger than they are.

@@kurisu7885 as someone on Role-play servers my hands are tied until they address this just to be safe. I’m also in a couple r34-esque servers but everyone there seems fine. I feel like this “false flag ban “ thing is VERY uncommon. No servers im in have posted about it which means no verified cases in any of them. Slight fear-mongering in the vid. But that doesn’t make any of this acceptable. Despite this update the problem remains. If they used ONLY photo DNA and a human reviewed THOSE flags things would be fine. This is what i would want. Photo DNA is very accurate. But the fact it’s combined with other so systems causes issues

@@fadeddragon123 Especially AI. I've seen some of the cursed stuff it can put out even with clear instructions.

that'd make it easier for bad people to share images and would most likely cause a ban from Google Cloud.

ehhhhhh, pornography of people who look younger

@@andoy_ Nudity*

@@SubtotalStar850-uh8pg The actress was fully clothed though.

@@leahnzastrzelecki5217 could have gotten someone else banned and then put her face in a database

@@leahnzastrzelecki5217 He was using that photo as an example for the video. It is safe to assume the person sharing images of her who got banned was sending unclothed ones.

Even 99% isn't acceptable, because imagine that the filter checks 1 picture every day: 99% to the 365th power is a whopping 2% chance that there will be no malfunctions.

Yep that's exactly what it means

Nothing new .. I had a discord account since 2019 I made on my PC. Last month it required phone number. Also lots of servers I've been using now want not only verified phone number, but also link to your KZfaq/Reddit/ steam account, Gmail and SMS to block "alts" ( I don't even understand why having alts is bad, never been banned ever but still sometimes people forget login)

@@LoveLaughlardlivelaund It's so people can't evade a well-deserved ban so easily. Makes sense when trolls appear en-masse.

because that would require a human to evaluate it, and that means the 'smol bean uwu tech company' would have to hire people/a firm to do it, which means they don't get to have as peachy of a bottom line. just pure cheapness.

They just need to pretend they're doing something. Investors don't use the platform,they just hear them say "we have systems for that in place" and go with it

Ya cos this is just about data exfiltration and doxxing

I think this is a bad take. How many messages a minute is sent through discord? How many languages, how many images, etc, etc. You ask for the impossible. There's a reason sites like Discord and KZfaq automate this stuff, and then use a human review as a second layer. Yes, they can still have half-cooked solutions but what you ask requires a literal army.@@NightfallGemini

I mean what do you want the admins to do, ban themselves?

WTF FRR??

most AI models will literally download everything available on the entire internet, and using it could be okay if it is going to prevent more of it in the future. Its not like there are discord employees with 400gb of cp on their pc is it?

They most likely get their training data from people who have already been banned for sending child material, not actively looking for it. But still, it's kinda messed up.

@@EnderReaper64 most training data is just whatever is available anywhere on the internet, even in the worse areas, thats where they get it from. Its also not like someone is actively looking for it, they would have a program to just automatically get all the data and it would be analysed by an AI, a human never sees it. Also im sure that for privacy concerns discord cant use your images or messages for AI training

​@@JoyousJoeydiscord is the only social media where deleting smth erases it fr without whizz

Remember, the height of consent is 5'5"

Even he most petite, flat chested woman still visually looks like an adult, sorry. There's just a look that young girls have. "Some teenagers look more mature" never seen it. Every one I've seen, I've clocked as a teen. Same with "adults who look like kids" you can TELL. You really can.

​@@jadecoolness101Maybe some people can tell. I can't speak outside my own experience. After I shave and get a haircut, I'm told I look like a high schooler (not directly, but people assume I am), which is years past for me. And as far as my personal, experience, I really can't. I've met women I thought were 12-14 but were older than me. Maybe you have magical eyes, and can tell better than the rest of us. But one way or another, don't assume everyone has a capacity to do something simply because you can.

@@jadecoolness101 I've seen a lot of women in person who look exactly like teenage girls, this problem is not one with a 'one size fits all just look at them really hard' kind of solution 🤷‍♂ this also doesn't account for women who have conditions that make them look like children

@@jadecoolness101 So you've never seen an adult with braces before? You've never seen a woman with a soft face before? Most 18 year olds don't look like they're in their late 20's dude that's just you

indeed my friend

So they can sell it to the Chinese 😂

never believe ad copy. every corporation is a soulless one, no exceptions.

​@@NightfallGeminidiscord wasnt a soulless company in 2016,just a bunch of guys

just like everything in life... the bigger and more famous it gets, the more it tries to squeeze money out of it, one way or another there are no exceptions, not even elon musk, he's full of shit as well

It scans dms, but it doesn't save the data or use it for training. Also, AI might not understand it word for word or like that, it would basically just search for keywords or phrases within messages, or aspects within photos.

No, I'm pretty sure there's no guarantee of privacy from Discord Inc. themselves. Probably in the terms of services somewhere. Or they'll just say they do but they don't, or have some rationale that'll justify it. Twitter's the same way, that's why they're "Direct Messages" instead of "Private Messages". Anyhow, you can't trust a centralized, hosted service like Discord to be completely private. If any of us _really_ valued privacy, we'd be moving to self-hosted services and software en masse, like Matrix/Element.

​​​@@SwordfighterRed Exactly, nothing is truly private if its going through a third party's machine

@@crebz i DM'd with their dumb chatbot and its pretty easy to verify that it remembers information that you send it. who's to say discord doesn't do the same?

@@crebz Soo you agree that all your data and pictures sent in discord can be seen by an entity as long as its 'not saved'? You might as well let cops search your home and devices since they won't save it, but arrest you afterwards for bogus claims. This is where discord is heading with these AI and random ban waves

I think getting banned means they want you to quit

Same

@@playtimeplay4518Murderer Zionists also want to end all online human communication…

Just don't send weird shi and don't send stuff that clearly labels as something that will get your account banned and you'll be fine

​@@KaxologyDepends on what discord consider such and if they keep to add random shit to those terms

No human at Discord is even smart enough to think of that, much less actually do it without making every Discord user on Tumblr and some of the ones on Twitter mad enough to organize a day to all request their data out of anti-company spite.

@@sleepy-emerald Honestly, you’re right. If they were actually smart, they would have done some great things with Discord.

You didn't watch the video

@@psgamer-il2ptYep, didn’t have enough time since right after, I got a fever and have been dealing with it. So now I looked and stand corrected. Thanks for pointing that out

tOo eXpEnSiVe NeEd ToO oUtSoUrCe ThIs sHiT bet what discord is thinking while literally having millions

Yes blame the users for Discords shitty system. Great idea.

innocent people always suffer thats why humanity has wars for centuries

@@terrycrews1584 yes, blame the malicious users, not the innocent ones... What is this sarcasm bro?

@@ZiarayZ As we could see, you dont have to be malicious to get caught by the system.....

there's a reason why mods on the world's biggest adult video site have ptsd

hey can i eat cheese pizza

It's still happening.

or child porn, as it is the actual meaning

@@worthaquintillionit’s the same thing it’s just a diff way to say it

Hahaha, that's on top of the case backlog for everything else. So good luck getting in front of a judge in the next 2 to 5 years. At the very least, it's gonna get some very bad people off because of due process violation defenses. Your right to a speedy and fair trial just got nuked because some politicians got touched by the good idea fairy.

In fact they actually kinda violate them more by looking at their images without their consent

@@SubtotalStar850-uh8pg The images are sent to the correct law enforcement agencies and they do the checking. No one else is allowed to legally view those.

ur dumb if u think disocrd looking at it when the government is obviously the one who make the laws about this shit@@SubtotalStar850-uh8pg

@@SubtotalStar850-uh8pg Any anyone's images who gets falsely flagged, heh

But they allow loli if it's furry for some reason

​​​@@SubtotalStar850-uh8pgDiapercub is still explicitly not allowed in the tos, what do you mean?

initially it was not. @@NoctisTheBogWitch When they first announced the ban, cub was exempt and people were outraged and started doxing employees. they found that one of them was a known cubfur so there were huge accusations of bias that didn't stop until the rule was amended.

@@peepysprite Do I look like I'm fuckin stupid or something? I knew that.

@@NoctisTheBogWitch You asked what he meant. Rude.

>not even no no types then I doubt that that was the reason. humans can be stupid but not stupid enough to look at a clothed person and go "yes, this is a naked child"

exactly I also got banned today for no reaosn

speaking of getting banned, I just got banned yesterday for the sole reason of being in a porn server and I VERY RARELY INTERACT IN THERE

​@@iorinorimaki3800Then don't be in risky servers? You did this to yourself.

@@NoctisTheBogWitch lol. "risky servers" are any server that a bot can join and spam images in. put the blame on the guys who run the platform, not the people using it.

My favorite is how the US military tested an AI to detect movement, it accurately spotted people walking and running, but completely failed to detect somersaulting, hiding in a cardboard box, and one particularly creative guy who held out sticks in his arms and convinced it he was a tree. These models are extremely easy to fool, and people have recently discovered that they can also be tricked by adding inaccurate descriptions to photos as metadata.

@@asteroidrules Fuck me, I forgot all about the army guys fooling the robot. I think the funniest part of that is that the report in question goes out of its way to point out that the soldiers were giggling the entire time. They _knew_ it was Looney Tunes shit. If you invented a robot whose purpose was to detect people, and it fails to notice a grown man playing the best tree in the school Nativity play, that's pack-up time. You somehow misappropriated your grant money. So you tell your robot to look for soldiers people holding branches. And next, it's going to overlook a soldier approaching on a bike. Look out for that, and the next soldier is going to be approaching on his belly like a snake. Look out for that, and the next soldier is going to be dressed in a mascot costume... As the scope of an AI's responsibility widens or the criteria it has to use expands, the amount of data it needs to carry it out increases, and it's going to bloat. An intelligent robot is never going to be a good substitute for a moderately clever, neuroplastic human, but people are expecting it to do creative endeavours like art and poetry, or make serious decisions like what the military eventually plans to. I can barely get ChatGPT to tell me what word I'm thinking of, and techbros want it to punish CSAM and terrorism without being able to program into it the importance of things such as context or human errors.

@@Noxedwin Yeah, there was also the recent development of Nightshade, a tool that embeds misleading data into images in ways that humans can't see when viewing the image, but is immediately picked up by the scrapers used to train these AI systems. A tool like that can very easily fool an AI into either a false positive or false negative if it's known just what the AI is trained to detect.

It gets worse. If you have access to the classifier being used (i.e. the tool that says "it's an apple") you can make minute adjustments to the image to nudge in a direction you want. If your goal is just to make it say something different, you can almost certainly guarantee that you can do it without the image looking off. In other words, you would need a side-by-side comparison to notice any difference. If you want it to say something specific, it might be a bit harder, but you can probably still find an image that will be classified as whatever you want while looking like a normal picture of something else. One way of doing this is called "Fast Gradient Sign Method" if you want to look it up.

That's not how it works. You're thinking of image captioning models, not image classification.

Source?

As a victim of a two hundred member doxxing group, that had to be brought down by LAW ENFORCEMENT, yeah, my life would've been a lot better now if discord did their job.

you send nude pics of yourself on discord???

I'm pretty sure highschoolers are allowed on discord though

@@Yoyiiii they arent allowed to send nude pics

​​@@Yoyiiii some people have romantic partners they send nudes to lol

The average start of puberty is 10-12 so the kids already know

cue the AI bros appearing and saying "AI is a tool" as if it hasn't done enough damage already, or could ever cause even worse problems down the line. i'm holding out for the superintelligence to usurp the governments and either calculate how to solve our issues without killing us or just get it over with and vaporise us lmao i'd prefer option 1 but i'm also fine with 2

say what you want but I don't envy the mf that has to sit there and look at "Club Penguin" to determine if it's the case.

@@Kaxology Great replacement terminology. Rough subject, but the phrase made me chuckle

@@spacebassisttrue this is why we should ban all guns, cars, knives, anything that can be misused for crime or to hurt people

There are plenty of good uses for AI and bad uses. Only dumb or lazy people/corporations use AI like this.

Good now hackers can easily nuke your account if you get hacked none of this is good and you probably won't be able to get it back.

the block button exists, but yeah reporting sucks obviously, and its even worse if it was on other places too

At this point, any update of Discord is gonna get anyone paranoid.

get a life?

@@faetalize get a life?

@@Nyanderthall Get a life?

@@faetalizesorry sir, you must be mistaken, no one here can get a life

They don't exactly, they do math of known child stuff to represent the image as a bunch of numbers and then do the math to see if it matches. It's called PhotoDNA

@@ashleyberkowitz8772 the images are a bunch of numbers. Always are. All AI models represent the images as numbers. That's how the perceive the image.

You two are not correct, this would only apply to the completed model once its done training. But PhotoDNA processed the real deal to train a model or several, which yes, then becomes basically a tokenization of metadata aka you can't replicate anything from it besides the classification. It still required PhotoDNA to basically go through hundred thousands of illegal images, which is legal for big corporations to do because they get grants from the governments across the world to build these systems in the first place. They still do whenever something is reported, so it has to process and learn from new illegal images. Obviously this could be a controversial point when it's juxtapositioned with generative image models or ChatGPT, because like with other models, the training process used images without consent and potentially - if you were a victim - the state/government had given out illegal photos of you to help subsidize massive for profit corporations.

I remember coming across a yt video explaining AI art and "cheese pizza". Same type of stuff here with AI art.

@@34125867 that's exactly my point?

i'd say yes

saying that is hella suspicious in itself weirdo

@@ANALBONES I'm not saying there should be no attempts to combat CP on the internet. But when innocent people are ending up in the crosshairs again and again, then it goes too far. And I'll remind you that it's not just happening on discord.

yes. It's a distinctly American problem. Most other countries are very much less concerned about sex in general. Americans get offended and call out a woman for breastfeeding in public, meanwhile public nudism in other places is fine, and often a normal part of their culture. Its so bad in America that if your a single dad and you take your teenage daughter out for ice cream, some ppl will automatically assume your dating/grooming a child and violently intervene. (there's videos where this happens) The issue is everyone here is brainwashed to think about sex all the time and to always assume the worst.

Um, no

@@potatopathsofglorylol yes it would what use your brain

@potatopathsofglory how is that not discrimination, Mr. Potato sir

@junyaiwase it's a potato, it doesn't have one

​@@junyaiwasenot legally speaking

wellll theres a reason why those freaks at roblox eventually take their heinous conversations to discord

KZfaq and reddit are probably still in the lead

Yeah, this has always been a problem where moderation was automated. Especially noticeable in online video games

@@crepooscul And it's made so, so much worse by the complete lack of an appeals proccess, since that means you won't get the false positives fixed via appeals.

Rockstar games is even doing it and people are calling them out left and right for it.

While it seems lazy, you do have to understand that there are, quite literally, millions of users sending billions of messages. DAILY. The sheer amount of employees needed is not feasible. Especially if they are also paid a fair wage. And this isn't touching on the fact that most people don't want their DMs being personally monitored by some rando at discord. There should be an appeal process, with humans behind that. But day to day monitoring is not feasible using humans alone.

@@jadecoolness101 Ok, then only do this when something gets reported instead of monitoring everything constantly.

Garden snail? PHALLIC. CHILD PP.

lol been years if youre android tablet user

They already did this when anything of the sort happened. I know this because my friend was in a NSFW server which had a private section for child no-no material (which he wasn’t part of) and they just banned him and a bunch of other people that didn’t know any of this was going on. I highly doubt they will keep this new “bot” in place.

I still have my SKype account just in case, that said I don't know of other chat services that still operate.

I wonder what type of person would be attracted to such a position

Just like working in morgue, but that is also very necessary

That's not how that works, but ok

​@@infinitehexingtonyes it is, discord workers get to view any image that gets reported not cool with that

It 100% will be used for that purpose, maybe before the 2024 election.

you're literally a nazi

They are losing money probably and fired a lot of them. They have to do this. I bet this is why they push nitro so hard too.

@@Lusc1nt They are, and it has to do with the fact Discord is a bad program. The only thing that made it better was the fact it was better than Skype, which was the dominate chatting platform at the time. Teamspeak costs money and the quality is shit, no one sane uses it, MSN and Xfire were both the same, terrible quality and tedious to navigate. Skype came along and made it easier, annoying, but easier, then Discord came along and did what Skype did but a little better. Now Discord is sitting on its laurel pretending it's the king of all programs, and it's dying. If it wasn't the Discord staff being found in erp discords causing the purge, then it's discord staff being caught buying APIs for their servers prompting a discord response, if it isn't Discord staff being caught chatting with underage people it's them at the same time banning people who mention theyre underage. Discord is doing everything in it's power to not admit that 60% of its staff are pedophiles and exist in the company purely to erp with kids and clear the logs at the end of the day, if this wasn't the case, Discord would be one of the best free to use chat programs out there, but it isn't.

​@lordrevan571 wats worse is if u send furry stuff regardless of how it looks ur safe, but send an anime loli u would get banned

@@lolidemon3163 Vast majority of the staff at Discord are or were furries, so this tracks.

@@lolidemon3163 Send 1 UNDERAGE GIRL AND THE WHOLE WORLD FREAKS OUTTT

fr

yeah the tax year is over, so they want to indoctrinate new users and retain all those gifted nitros

This is why I deleted the app earlier, So fucking glad I did, Discord is just getting worse and worse.

​@@ChrisDaNoob76 what do you use instead

@@jc_art_ I used Reddit back then BEFORE I moved to Discord, But then I got back into Reddit because I feel alot better there then Discord (Quick note, I’ve barely used Social media apps so I’m not the biggest user of them)

Keep in mind you never will know if was no broken normal people don't look for illegal material there

Thing is, it is broke, but they failed to fix it. There's servers dedicated to porn, lolis, furries, etc which just keep working but random innocent people can get their privacy invaded and account removed for no reason at all now.

The false-positive rate of PhotoDNA is said to be 1 in 50 billion. That's as good as it can get. The issue with ML/AI is that the false-positive rates are never going to be that high, due to the sheer nature of it all. Even with basic stuff, it tends to get things wrong. Classifier tech is not designed to handle high-stakes matters.

@@sylvie7868 "machines should not be allowed to make managerial decisions" would have been the best rule to follow, really, but here we are