> In captions, ads enticed Meta users to click with claims like “I can show you more” or promises that “our AI transforms your hidden desires into hypnotic videos. Dark, seductive, unforgettable.” Another boasted, “there are no restrictions. All the characters are real people and there are many options. This is the AI men actually use.”
> Even more troublingly, TTP emphasized in a press release that these findings came after the group teamed up with Wired and flagged an earlier batch of 53 ads in August. At that time, Meta claimed that it had “recently rolled out new systems to better detect child exploitation.” But TTP “found hundreds of new CSAM ads ran after Meta boasted about its new technology.”
Prison is too good for the people responsible for this.
Garbage, too. I've reported at least 10 of these things. Ads, reels. Specifically as child abuse. "After review, we determined this content does not breach our policies."
To be really, really fucking clear, it's not just "failed to catch" it's, "even when reported, failed to do anything about".
I got ads, reels that were not-at-all veiled hints at CSAM. "Report: CSAM" "We reviewed your report and determined that the content does not breach our policy."
That's the rub. And I think that they've found my "well, fuck this" point. I already don't frequent their products, but I think it's time to disentangle what I do have on there from other services so I can download and delete.
Presumably these ads are targeted using the algorithm. Thought crime is scary and all that, but it's an interesting part of this system to consider. And then there's the question of these images living in your cache if you happened to be shown one. "Welcome to the internet.."
The "Kompromat" angle should not be ignored in this day and age.
"In cases of kompromat during the early 21st century, Russian operatives have been suspected or accused of placing child pornography on the personal computers of individuals they were attempting to discredit." https://en.wikipedia.org/wiki/Kompromat
The only real question is how easy it would be to target the undesired via the ad placement targetting options on social media platforms. On the one on which I've run ads (LinkedIn), pretty hard. At least on a technical level, because if your audience size is too small it asks you to expand your criteria. Maybe other platforms are less strict?
For sure would be a more comfortable approach for state operatives than physical access.
--
Aside. Watch the movie with the same name. Use an ad blocker, religiously.
I’ve got a Facebook account that doesn’t feature my real name, has no friends and follows no pages. The algorithm decided that I would most like to see CSAM, flat earth theory and Sovereign Citizen nonsense.
Surprised it hasn't been closed already. Sometime back before I left entirely in 2017, I tried making an account that wasn't tied to my real name, email, phone number, etc. and after a few days they asked for proof of ID because they couldn't tie me to their "social graph" or whatever. Didn't give ID and they disabled the account.
That's interesting. I've noticed similar things on youtube. It seems like by blocking most of the tracking, it automatically falls back to AI slop and wellness gurus. I've tried watching some of these ads out of curiosity and it's often difficult to even figure out what the actual product is. It feels very scammy. Certainly light years from the early promises of google offering text-only ads when everyone else had malware laden banners and toolbars.
The only things I can think of that have pushed these into my viewport are that I follow multiple AI news pages, and multiple photography pages, same with my IG.
>the Tech Transparency Project (TTP) reported that Meta failed to detect 332 ads containing CSAM this year
Considering the amount of ads they handle that seems... reasonable?
Not to mention the stupidity of calling the result of nudify apps "CSAM" when no minors in situation of sexual abuse are used in any way to generate those photos.
> Considering the amount of ads they handle that seems... reasonable?
Why reasonable? Don't these ads pass through pre-publication review, and Meta makes money off of it? Any number above zero is a serious problem. If they can't run an ad business that doesn't publish and profit off of CSAM, they shouldn't run it.
I don't think we should distinguish between 'fake' and real CSAM based on if kids are hurt; this ad machine shouldn't be able to publish it, regardless. The moral question of if it's real or kids were directly harmed seems separate to me.
I recall YouTube was hijacked to show multi-hour ads during the pandemic. These are imperfect systems that can be abused at any scale.
They identified them after the fact and appear to have disclosed that. This is what they needed to do to be able to be held accountable and to do better detecting them in the future.
I block ads and ad services at the device and network level and don't use any Meta services (outside of facebookresearch models), so while I'm fortunate not to have been able to be served any of these, it's also kind of unfortunate that I'm in a relatively rare category to know to report them and how.
Without more context it’s hard to say whether 332 is reasonable, but I interpreted “reasonable” as in there will always be some amount of harmful content uncaught by automated systems.
In my mind there are two components of this that are in direct tension though, regardless of numbers:
1. Humans having to review ad content is legitimately traumatizing and awful
2. Automated systems can miss obviously harmful content, which should never be allowed
To me this means we probably shouldn't permit these systems to scale as they do, because the company profits from harms to its own workers and/or ad viewers. I worry that we've normalized a business that shouldn't exist in its current form, but it prints so much cash that no one is willing to stop it.
> The young royal’s image was “animated into a video of her performing a graphic sex act,” TTP found. Other ads animated a photo of a 14-year-old Instagram influencer “showing off her new sports club uniform” into “a video of her performing oral sex.” A third “preteen” victim “posing in a pink athletic outfit with pigtails” in a series of stock photos was morphed into a video where she looks frightened as she’s molested by an adult male, TTP reported.
This is one of those topics that gets radioactive quickly, so I'm trying to parse this comment under the recent ruling by a judge that certain obscene products of generative AI are protected under the First Amendment.
"We now live in an age where GenAI [generative AI] models can render images depicting the abuse of virtual children that are virtually indistinguishable from those depicting the abuse of actual children. This case illuminates how this evolving technology complicates the lines drawn by the Supreme Court in Stanley, Osborne, and Free Speech Coalition. Indeed, in Free Speech Coalition, the Supreme Court addressed the scope of First Amendment protections for virtual CSAM, but that was nearly twenty-five years ago, and the image-generation technology available today was likely unimaginable back then. Given the relentless advancement in artificial intelligence models, we have some concerns about the lines these cases draw, but we are not free to redraw them ourselves."
Despite the source images not being CSAM, the products of nudify services may be CSAM but not child pornography. Because nudifying minors involves generation of material intended to depict real people who cannot give consent on computers outside the home, I'd consider them not to be protected speech.
> In captions, ads enticed Meta users to click with claims like “I can show you more” or promises that “our AI transforms your hidden desires into hypnotic videos. Dark, seductive, unforgettable.” Another boasted, “there are no restrictions. All the characters are real people and there are many options. This is the AI men actually use.”
> Even more troublingly, TTP emphasized in a press release that these findings came after the group teamed up with Wired and flagged an earlier batch of 53 ads in August. At that time, Meta claimed that it had “recently rolled out new systems to better detect child exploitation.” But TTP “found hundreds of new CSAM ads ran after Meta boasted about its new technology.”
Prison is too good for the people responsible for this.
Garbage, too. I've reported at least 10 of these things. Ads, reels. Specifically as child abuse. "After review, we determined this content does not breach our policies."
Expecting a corporate to police itself is, obviously, a joke
Of course. No large company cares about child abuse. They've all been to Epstein island, or one of the other unknown islands.
Hah, I don't know if Zuck has visited, but if he or any of the exec team has, you can bet that filter is finely finely tuned.
To be really, really fucking clear, it's not just "failed to catch" it's, "even when reported, failed to do anything about".
I got ads, reels that were not-at-all veiled hints at CSAM. "Report: CSAM" "We reviewed your report and determined that the content does not breach our policy."
That's the rub. And I think that they've found my "well, fuck this" point. I already don't frequent their products, but I think it's time to disentangle what I do have on there from other services so I can download and delete.
Presumably these ads are targeted using the algorithm. Thought crime is scary and all that, but it's an interesting part of this system to consider. And then there's the question of these images living in your cache if you happened to be shown one. "Welcome to the internet.."
The "Kompromat" angle should not be ignored in this day and age.
"In cases of kompromat during the early 21st century, Russian operatives have been suspected or accused of placing child pornography on the personal computers of individuals they were attempting to discredit." https://en.wikipedia.org/wiki/Kompromat
The only real question is how easy it would be to target the undesired via the ad placement targetting options on social media platforms. On the one on which I've run ads (LinkedIn), pretty hard. At least on a technical level, because if your audience size is too small it asks you to expand your criteria. Maybe other platforms are less strict?
For sure would be a more comfortable approach for state operatives than physical access.
--
Aside. Watch the movie with the same name. Use an ad blocker, religiously.
Do you mean this?
https://imdb.com/title/tt13349892/
Yes that is the one. I'm surprised by the low rating on IMDb. Well produced movie, with an interesting real-life inspired premise, a bit of thriller.
I’ve got a Facebook account that doesn’t feature my real name, has no friends and follows no pages. The algorithm decided that I would most like to see CSAM, flat earth theory and Sovereign Citizen nonsense.
Surprised it hasn't been closed already. Sometime back before I left entirely in 2017, I tried making an account that wasn't tied to my real name, email, phone number, etc. and after a few days they asked for proof of ID because they couldn't tie me to their "social graph" or whatever. Didn't give ID and they disabled the account.
That's interesting. I've noticed similar things on youtube. It seems like by blocking most of the tracking, it automatically falls back to AI slop and wellness gurus. I've tried watching some of these ads out of curiosity and it's often difficult to even figure out what the actual product is. It feels very scammy. Certainly light years from the early promises of google offering text-only ads when everyone else had malware laden banners and toolbars.
The only things I can think of that have pushed these into my viewport are that I follow multiple AI news pages, and multiple photography pages, same with my IG.
"Not sorry, it will happen again."
If only they were as liable for this kind of stuff as they are for DMCA'ble content lmao
>the Tech Transparency Project (TTP) reported that Meta failed to detect 332 ads containing CSAM this year
Considering the amount of ads they handle that seems... reasonable?
Not to mention the stupidity of calling the result of nudify apps "CSAM" when no minors in situation of sexual abuse are used in any way to generate those photos.
> Considering the amount of ads they handle that seems... reasonable?
Why reasonable? Don't these ads pass through pre-publication review, and Meta makes money off of it? Any number above zero is a serious problem. If they can't run an ad business that doesn't publish and profit off of CSAM, they shouldn't run it.
I don't think we should distinguish between 'fake' and real CSAM based on if kids are hurt; this ad machine shouldn't be able to publish it, regardless. The moral question of if it's real or kids were directly harmed seems separate to me.
I recall YouTube was hijacked to show multi-hour ads during the pandemic. These are imperfect systems that can be abused at any scale.
They identified them after the fact and appear to have disclosed that. This is what they needed to do to be able to be held accountable and to do better detecting them in the future.
I block ads and ad services at the device and network level and don't use any Meta services (outside of facebookresearch models), so while I'm fortunate not to have been able to be served any of these, it's also kind of unfortunate that I'm in a relatively rare category to know to report them and how.
Without more context it’s hard to say whether 332 is reasonable, but I interpreted “reasonable” as in there will always be some amount of harmful content uncaught by automated systems.
I agree, more context would be helpful.
In my mind there are two components of this that are in direct tension though, regardless of numbers:
1. Humans having to review ad content is legitimately traumatizing and awful
2. Automated systems can miss obviously harmful content, which should never be allowed
To me this means we probably shouldn't permit these systems to scale as they do, because the company profits from harms to its own workers and/or ad viewers. I worry that we've normalized a business that shouldn't exist in its current form, but it prints so much cash that no one is willing to stop it.
I'm tired of the argument that since it would be impossible to police at scale then they shouldn't be liable.
Just accept money for CSAM ads without accountability, it's fine. No ethical issue here. Why is everyone so angry? /s
How many ads for CSAM do your local TV networks broadcast? How many billboards do you drive past featuring CSAM? How about newspapers?
If a company can't operate within the law due to the scale it chooses to operate at, then it should choose not to operate at that scale.
It doesn't become acceptable to break the law because your scale means there a lots of times you don't break the law either.
> The young royal’s image was “animated into a video of her performing a graphic sex act,” TTP found. Other ads animated a photo of a 14-year-old Instagram influencer “showing off her new sports club uniform” into “a video of her performing oral sex.” A third “preteen” victim “posing in a pink athletic outfit with pigtails” in a series of stock photos was morphed into a video where she looks frightened as she’s molested by an adult male, TTP reported.
[flagged]
This is one of those topics that gets radioactive quickly, so I'm trying to parse this comment under the recent ruling by a judge that certain obscene products of generative AI are protected under the First Amendment.
"We now live in an age where GenAI [generative AI] models can render images depicting the abuse of virtual children that are virtually indistinguishable from those depicting the abuse of actual children. This case illuminates how this evolving technology complicates the lines drawn by the Supreme Court in Stanley, Osborne, and Free Speech Coalition. Indeed, in Free Speech Coalition, the Supreme Court addressed the scope of First Amendment protections for virtual CSAM, but that was nearly twenty-five years ago, and the image-generation technology available today was likely unimaginable back then. Given the relentless advancement in artificial intelligence models, we have some concerns about the lines these cases draw, but we are not free to redraw them ourselves."
https://www.snopes.com/fact-check/judge-ai-child-sex-abuse-r...
Despite the source images not being CSAM, the products of nudify services may be CSAM but not child pornography. Because nudifying minors involves generation of material intended to depict real people who cannot give consent on computers outside the home, I'd consider them not to be protected speech.
> Considering the amount of ads they handle that seems... reasonable?
Every time they disseminate one of these ads, they commit a serious criminal offence. What threshold of criminality is reasonable to you?