Post

NEW: In one of the most horrifying stories I've ever reported, I found that Facebook is hosting huge quantities of AI-generated videos of violent child abuse -- kids being beaten, burned, and far more But the genuinely shocking thing was how little Meta seems to care futurism.com/artificial-i...

719 reposts1.6K likes

I shut down my Facebook page about a year ago because my feed was suddenly filled with AI-generated photos of graphic animal injuries. I complained and they said it didn't violate their rules. They don't give a fuck about anything. Everyone needs to leave that cursed site.

That's terrible shit and very disturbing. There have always been unsavory characters on the Net but AI is making it easier for people to create and distribute the material. Platforms like FB need to remove and restrict the posting of material but it comes down to stopping those posting it with laws.

Are there patterns in how this is targeted? eg are the videos presented as 'true crime' stories, or 'kids have it too easy', or 'women are secretly monsters'. or targeting specific demographics? Or are they just spammed indiscriminately by pure engagement farming algorithms?

Maybe they “aren’t intentionally getting it wrong” Zuckerberg’s instinct was to allow Holocaust deniers to stay on. Took a few years more to kick them off. He just doesn’t care. More users, more time, more money. Move fast and break things

I've witnessed this same type of automated hate first hand. Facebook doesn't give a fuck,they're making money they so desperately need because zuck can't stop blowing billions on ideas only zuck likes bsky.app/profile/dany...

coming soon to an ai server farm near you: ai videos of paid protedtors with colored hair crying about their food stamps getting "snatched" away by "the recent cuts" by "former president trump". this is all the evidence i need to know ebt isnt being abused, or else they wouldnt need ai for fake vids

because folks in nazi cults think violating all societal norms is transcendental, more or less. they think any kind of morality at all is a weakness so they try to purge it out of themselves by doing this kind of thing.

Having lived a childhood much like that, I cannot tell you how much it just does not matter that those kids aren't real. Facebook / Meta / Zuckerberg feeding child abuse fetish fantasies will absolutely result in real-world harm - and it's got me feeling super vigilante-ish right now.

It felt to me like the big question facing Meta is: does this content have any place on its platform? But the general sentiment I got from the company boiled down to "please stop bothering us"

We conducted a test, by using regular Facebook accounts to report a bunch of accounts sharing the horrifying AI-generated child abuse content The outcome? Meta's moderators told us that the majority of the accounts don't "go against our Community Standards"

Not as disturbing, but I saw similar with Reddit suddenly getting a bunch of vids of absolutely decrepit old animals that I swear were fake (in some cases I'd had those species at that age and they weren't arthritic and mangey). Baffling that people would fake THAT, but I guess suffering get clicks

Interesting line from Meta's PR team: “Not all the links you sent violate our policies,” it read. “Can you please avoid suggesting in your reporting that they do?” I don't think the underlying question is whether the content violates Meta's policies. It's why doesn't it?

No, the shocking thing is that the Trump-Roberts SCOTUS has ruled that AI generated porn and abuse is perfectly fine. The Family values conservatives running & using facebook !

I routinely report Facebook groups & the posts in such groups to the Meta "Community Standards team", which is nothing more than an AI generated survey. It doesn't allow for submitting screen shots either. I've reported hate groups where Islam, Catholics, Jews, immigrants, racial minorities &

members of the LGBTQ+ community were targets of blatant hatred, wishing ill-will, including veiled death threats against these targets. Meta's response comes within a few hours with a pre-written response saying that they don't see it as a violation of Community Standards.

They have been doing it for a long time now but nothing has been done other than revelations never really acted on. The difference is that before AI was used it was actual children.