Фэйсбүүк платформ дээр хүүхдийн амь насанд аюултай, зүй бус үйлдлүүдийг харуулсан хиймэл оюун ухааны аргаар бүтээсэн бичлэгүүд олноор нийтлэгдсээр байгааг мэргэжилтнүүд анхаарууллаа.
Өнгөрсөн сард Фэйсбүүк сүлжээнд хүүхдийг зодож, шатааж, торны цаана хорьсон агуулгатай хиймэл оюун ухааны бичлэгүүд их хэмжээгээр илэрсэн талаар мэдээлсэн. Эдгээр бичлэгүүдийн дунд нярай хүүхдийг зууханд хийж байгаа дүрслэл зэрэг хүнлэг бус, аймшигтай агуулга бүхий дүрс бичлэгүүд багтаж байна. Тус компани эхэндээ зарим дансыг устгасан боловч удалгүй тэдгээрийг эргүүлэн сэргээж, агуулгыг хянах тал дээр тодорхойгүй хариу өгчээ.
Meta компани хиймэл оюун ухааны хөгжилд олон зуун тэрбум доллар зарцуулж байгаа хэдий ч өөрийн платформ дээрх аюултай контентыг илрүүлж, устгахад энэхүү нөөцөө ашиглахгүй байна. Тус платформын алгоритм нь бичлэгүүдийн тайлбар хэсэгт бичигдсэн хүчирхийллийн нарийвчилсан мэдээллийг хялбархан таних боломжтой байтал идэвхтэй арга хэмжээ авахгүй байгаа нь шүүмжлэл дагуулж байна. Өчигдөр 19 шинэ дансны агуулгыг шалгуулахаар илгээхэд тус компани ямар ч хариу өгөөгүй бөгөөд бичлэгүүд одоог хүртэл олон нийтэд нээлттэй хэвээр байна.
Энэхүү асуудал нь зөвхөн бичлэгээр хязгаарлагдахгүй бөгөөд өмнө нь тус платформ дээр бодит хүүхдүүдийн садар самуун зураг бүхий сурталчилгаанууд нийтлэгдэж байсан нь тогтоогдсон. 13-аас дээш насны хэрэглэгчдэд нээлттэй тус сүлжээ нь агуулгын зохицуулалтын тал дээр хариуцлагагүй хандаж байгаа нь олон нийтийн анхаарлыг татаж байна. Хиймэл оюун ухааныг ашиглан ийм төрлийн хор хөнөөлтэй контент түгээх явдал өдөр бүр нэмэгдэж байгаа нь Meta-гийн аюулгүй байдлын бодлого хангалтгүй байгааг харуулж байна.
Дэлгэрэнгүйг эх сурвалжаас харах
↓Эх сурвалжийг нээх ↓
Content warning: this story contains descriptions of disturbing AI-generated videos that depict violent child abuse.
Last month, we reported that Facebook was hosting enormous numbers of graphic AI-generated videos of physical child abuse that showed kids being beaten, burned, dragged by their hair, locked in cages, and more.
In response to our questions, Facebook’s parent company Meta alternately dragged its feet, rationalized about why the videos were acceptable, and issued hard-to-parse statements that dodged any accountability or meaningful commitment to fighting the phenomenon.
As we reported the story, Meta eventually pulled down an account we shared with it that had uploaded a particularly disturbing array of AI videos that depicted children being viciously beaten and forced to eat trash, as well as many more appalling scenarios including a young girl being locked in a freezer by a kidnapper.
But after we published our piece, we were surprised to see that the account had been reinstated, along with its sickening cache of videos. We reached out to Meta to ask why the account was live again and received a curt response from a spokesperson.
“Just wanted to let you know that we’re looking into this,” it read. “I’m not sure of an exact timeline, but I will let you know when I learn more.”
The account later vanished again, and the spokesperson ignored follow-up questions about why it had been restored to the platform.
Untold thousands of similar videos remain on Facebook, and more are being uploaded every day. Curious whether Meta had grown more responsive to reports about the distressing content, yesterday we sent it 19 more accounts loaded with similar child abuse horrors.
This time the company didn’t even reply, leaving all the accounts live on Facebook more than 24 hours later when we published this story.
The videos, which we are opting not to share, contain a cavalcade of horrors.
One shows a baby roasting in an oven, sobbing while its skin blisters from the heat. Another shows a toddler being pushed down the stairs. Another shows a little girl tied to a tree while a woman strikes her with a cast iron pan. Many more clips show children being beaten with belts, ropes, and other household objects, or being locked inside washing machines and dog kennels.
“Grandpa, I’m so hungry,” pleads a girl who’s been shut in a pigsty in one of the videos.
Another subgenre of video being posted by the accounts we sent Meta depict Black children being beaten and berated by teachers in classroom environments.
“Why don’t you go to a school for Black kids?” a teacher demands in one of the videos, after striking a crying Black student with a ruler. “Why do you keep trying to get into a school like this?”
(Many of the clips end with an adult stepping in to help the child, which Meta pointed to as justification for the videos in our last story.)
If Meta was serious about proactively identifying and removing the videos, it would have immense resources at its disposal to do so. The company has poured hundreds of billions of dollars into developing what its CEO Mark Zuckerberg terms “superintelligence” — AI resources, evidently, that aren’t being marshalled to control the spread of horrifying content on the company’s platforms.
In fact, the company could easily flag the child abuse videos simply by looking at the text descriptions that they’re being uploaded with, which often describe the torturous scenes in disarming detail.
“My mother put my newborn, who was suffering from a severe fever and chills, into a heated oven to ‘cure the cold,’” reads the caption on the video of the baby in the oven. “Even as the baby screamed and suffered severe burns, she refused to take the child out because she believed her treatment was about to work. When my husband rushed in to rescue our baby, she began smashing things, screaming, and blocking the oven, refusing to let him take the child out.”
The videos we identified aren’t the only AI-generated nightmares on Meta’s platforms. Last month, Wired reported that the company had allowed hundreds of advertisements that included overtly sexual material of real children.
Add it all up, and Meta — which allows users as young as 13 — feels like a company that’s effectively given up on content moderation. If anything, it just seems annoyed that we keep pointing out its complicity with the horrific content.
More on Meta: Meta’s Muse Is Spying on All Your Friends and Family
The post Meta Is Refusing to Remove AI-Generated Facebook Videos of Graphic Child Abuse, Including a Baby Roasting in an Oven appeared first on Futurism.
