Sabitlenmiş Tweet
Just got my new model, hopefully I'll be able to show it off on stream soon. It's been a long time coming
🎨: @maddnesswriter
#Vtuber #VTuberUprising #ENVtuber
English
Blackbluue
696 posts
Blackbluue
@BlackbluueVT
He/Him Model 🎨: @maddnesswriter Pfp: @_schlen_ Banner: @maddnesswriter EN Only
Wandering the void... Katılım Ağustos 2022
269 Takip Edilen73 Takipçiler
@BlackbluueVT I know, that's why I hate it. It's what got me signed out of Discord FIFTEEN TIMES. Nothing else does that
I mean my phone, being android, tells me that I'm low on storage and that things may run slower, but it doesn't completely break down like Windows
English
Why is Winblows so self-destructive when you have low storage
Apple- "Hey, you have low storage"
Android- "Hey, you have low storage"
Windows- "Goodbye cruel world"
English
@DrumstickPony Except AI didn't do any of those things in this instance.
English
This is what happens when you let AI be judge, jury, and executioner
AI should never ever be allowed to moderate and this is why
Discord Support@discord_support
This weekend, our safety systems incorrectly triggered and banned around 200 accounts. Everyone affected has been reinstated. Here's what happened. A thread 🧵
English
I've seen a lot of people quote this post, falsely claiming that Discord's AI moderation is at fault. Please understand people, this is not related to their AI moderation. it's a file hash false positive. Don't be a boy who cried wolf; attribute the correct problem to the issue.
Discord Support@discord_support
This weekend, our safety systems incorrectly triggered and banned around 200 accounts. Everyone affected has been reinstated. Here's what happened. A thread 🧵
English
@CCShakeCat @OzwaldV001 which is fine. as long as you understand what you are being careful of
English
@BlackbluueVT @OzwaldV001 I'm still being careful
English
And my friends really thought it was complete BS
Be very careful everybody
JDBRYANT 🎂 TODAY@jdbryantdev
My account was wrongfully banned from your platform due to a bug in your AI automod detecting my GAME TEXTURES as CSAM. I need my account back as I'm a game director and use Discord for all my communication. I have requested a review of my suspension. @discord @discord_support
English
Blackbluue retweetledi
This weekend, our safety systems incorrectly triggered and banned around 200 accounts. Everyone affected has been reinstated.
Here's what happened. A thread 🧵
Tall Cow@tallcowyt
There’s currently a vulnerability in Discord’s AI moderation that detects any and all square grid images, such as spreadsheets, chessboards, Minecraft inventories, white & gray transparent backgrounds… as CSAM and will permanently ban your account. To keep your account safe, I would recommend NOT sending any sort of images anywhere until this vulnerability is fixed. If you are a server owner or admin, I would highly recommend to disable all image and gif/embed perms in your server for the time being. To @discord @discord_support, this vulnerability needs to be fixed immediately with all false bans overturned. Additionally, AI moderation needs to be scrapped and replaced with humans. This is exactly why companies shouldn’t be using clankers to decide whether accounts should be banned or not. Losing a Discord account to something as unfair as this can be extremely devastating and affect users severely, and every day millions of users are affected by false AI bans. This needs to be stopped. If you would like to spread awareness about this, feel free to like & repost.
English
@weldon1045 can to state the relevance of this image? where was a computer ever given direct management decision power?
English
@ScrubaatiVidya but this isn't AI related. you are conflating 2 unrelated problems.
English
When will discord understand letting AI handle bans on that leave is just not good BECAUSE of this?
False positives are why I’m still convinced my account got false banned over something I’m still sure was not csam
Discord Support@discord_support
This weekend, our safety systems incorrectly triggered and banned around 200 accounts. Everyone affected has been reinstated. Here's what happened. A thread 🧵
English
@ChiiFett machine learning may or may not be a problem, but please don't state that this particular issue was AI related. it had nothing to do with AI.
English
Oh look yet another example of how replacing people with machine learning doesn’t work out. Remember the layoffs Discord had a while back!?
I use Discord every day. It is the life line to my online friends and my community. Please do fucking better.
Discord Support@discord_support
This weekend, our safety systems incorrectly triggered and banned around 200 accounts. Everyone affected has been reinstated. Here's what happened. A thread 🧵
English
@CoolausdtinDT people keep citing AI when talking about this, but this has nothing to do with AI. it was a false positive hash comparison. it's just comparing 2 numbers.
English
Here's an idea, don't use AI to moderate account activity. This is blatant proof that it's unreliable without a human being taking full control.
Discord Support@discord_support
We know that's not a satisfying explanation if this was your account, and we should have caught this sooner. We're working on better safeguards so this can't quietly happen again, and more broadly, on making sure our safety systems don't penalize people who did nothing wrong. If something still feels off with your account, reply here.
English
@StarcatTails also i could be wrong but possessing CSAM imagery of any kind in any capacity is then another broken law. so them storing and sending you back a clip of the offending image would cause both you and them to break another law.
English
@StarcatTails while i do agree with your sentiment, i just want to point out that this discord grid thing isn't related to AI at all. the issue comes from a photo hashing algorithm owned by microsoft.
English
With cases such as this with false AI bans, mass report bans, YouTube bans and copyright issues.
I would like to see a consumer protection law that any sort of ban, suspension or demonization must be done by human review, and they must provide the clip or content that caused the suspension.
I have seen so many get vague suspensions, demonetizations, I regularly have to deal with misidentified audio copyright claims because the AI misidentifies songs from the same artist.
Social media companies are protected from user speech, why is there no equivalent strongly enforced law to protect the end users from false ai flag, fraudulent, and vague bans and suspensions?
Dexerto@Dexerto
Discord has responded after users were banned for child safety concerns for posting harmless grid images like Minecraft inventories A developer says a 'faulty content hash' caused false child safety matches, not AI
English
@dootskyre @panley01 discord.com/safety/how-dis…
they've already been very public about using PhotoDNA to flag images though.
English
To the people screaming "AI!!", this is something worse.
Discord uses PhotoDNA fuzzy hashing to identify CSAM. One innocuous photo, one innocent frame in a video, that looks just enough like your upload?
Bam.
PhotoDNA is the industry standard & NCMEC endorsed. But... 🧵
Discord Support@discord_support
This weekend, our safety systems incorrectly triggered and banned around 200 accounts. Everyone affected has been reinstated. Here's what happened. A thread 🧵
English
@Dovahkiin1060 @cori6487 by "bot" i hope you are not claiming it to be AI. yes discord's code is broken. but it's not an AI bot banning people. it's a false positive hash collision produced by PhotoDNA, which is a microsoft product. discord has no control over that. discord just messed up the appeals.
English
@BlackbluueVT @cori6487 The fact that a ton of accounts are still getting restricted due to the rampant false violations given out by their broken bot, which they’ve falsely claimed to have fixed, lmao.
English
"which is why a member of our Trust & Safety team always reviews flagged content before any action is taken"
so we're just lying now?
Discord Support@discord_support
Our systems flag content by matching it against known harmful material. This kind of similarity matching can produce false positives, which is why a member of our Trust & Safety team always reviews flagged content before any action is taken. The intended behavior is to temporarily pause uploads during that review, not ban the account.
English
@BiggerDinger @SavageWuff yes, it can and should be improved, but it's not a simple matter of "changing the hash of a single image"
English
@BiggerDinger @SavageWuff more complexity means less collisions, but hashing takes longer and you get less accurate results when comparing 2 images. less complexity means stricter comparisons to better catch slightly different images and faster hashing, but as we've seen more false positives.
English
Discord core thread🧵:
>Devs lying about exploits not existing.
>Devs lying about human mods reviewing CSAM bans before applying them.
>lied about removing those bans ( grid/spiral ) exploit ones.
>exploits still exist even though i reported them 5 months ago including grid one.
English
@BiggerDinger @SavageWuff I did see this video back when it was released 2 years ago. and he directly says it's not AI banning people. my whole thing this entire day was to stop blaming AI for things AI didn't do.
English
@BlackbluueVT @SavageWuff also last thing
youtube.com/watch?v=Kyc_ys…
about the image thing like i said its related to photodna like ntts said too
YouTube
English
@BiggerDinger @SavageWuff the photodna isn't a discord issue. they don't own photodna, it's an industry standard. yes, it needs to be fixed. but not by discord.
English
@BlackbluueVT @SavageWuff for images like i said its related to photodna
discord gotta find a fix for it though cause grids and spirals not the only thing that triggers it
English