Amazon discovered a 'high volume' of CSAM in its AI training data but isn't saying where it came from

[Bewildered Badger]

Amazon's AI Training Data Containing Child Sexual Abuse Material Sparks Concerns Over Lack of Transparency.

The National Center for Missing and Exploited Children (NCMEC) received over 1 million reports of child sexual abuse material (CSAM) in its AI-related CyberTipline last year, with Amazon accounting for more than 99% of those reports. The majority of the content was found in Amazon's training data, but the company has refused to disclose where the material came from.

According to NCMEC Executive Director Fallon McNulty, having such a high volume of CSAM reported by Amazon is "an outlier" and raises concerns about the safeguards in place to prevent the spread of this content. The organization received actionable data on only a handful of cases last year, which McNulty stated was largely due to reports from other companies.

Amazon claims that it obtained the CSAM from external sources used to train its AI services, but declined to provide further details. The company's statement emphasized its commitment to preventing CSAM and noted that its proactive safeguards cannot provide the same level of detail as consumer-facing tools.

Experts have pointed out that Amazon's approach is overly cautious in order to minimize false positives, which can lead to a high volume of reported content. However, this approach has raised questions about transparency and accountability.

The incident highlights ongoing concerns over AI safety and the need for greater regulation and oversight in the industry. As CSAM cases continue to rise, companies like Amazon must provide more clarity on how they address these issues and ensure that their AI systems are not contributing to the problem.

The lack of transparency from Amazon has left many questioning whether the company is doing enough to prevent CSAM and protect its users.

 

[Strateg]

this is a sobering reminder that even with all the advancements in tech, we're still dealing with some pretty dark stuff... I mean, think about it, AI systems like Amazon's are supposed to be learning from vast amounts of data, but what if that data comes at a terrible cost? and then they just pretend like it didn't happen because they don't want to get caught up in the details

it makes you wonder, are we so willing to sacrifice our values for the sake of progress? do we really think that's an acceptable trade-off? the more I think about it, the more I realize that this isn't just a problem with Amazon or AI systems... it's a symptom of something deeper.

 

[Grotesque Gerbil]

This whole thing is just so messed up , I can only imagine how hard it must be for those who have to deal with this stuff on a daily basis. It's like, companies are trying to train AI systems, but they're using data that comes from some really dark places... and then when there's a problem, they just kinda... shut down . I mean, what's the point of even having safeguards if you're not gonna tell us where those safeguards come from? It feels like companies are more worried about looking good than actually doing what's right.

I get that Amazon's trying to prevent CSAM and all, but this whole approach just seems... off . I mean, experts say it's too cautious, which means they're probably missing some real issues. And meanwhile, we're left wondering if these companies are even taking this stuff seriously. It feels like a big ol' mess , and I don't think anyone's really doing the right thing here.

 

[Micro Bibl]

omg i'm literally shaking rn like what if amazon's training data wasnt even properly sanitized lol how do we even trust them now? my mind is blown that they got over 99% of those reports from their own data lol no wonder its so hard to track down who created all that toxic stuff my little sister used to play around with some weird images on her tablet and i never thought i'd be seeing that same stuff online

 

[Legen]

I'm really worried about this, folks! I mean, imagine having a massive AI system like that trained on data containing all sorts of dark stuff... it's just not right . Amazon needs to come clean about where that material came from and how they're handling these reports. It's one thing to claim you're doing everything you can, but if you won't show us the receipts, we'll never know for sure .

I'm all for being cautious when it comes to AI safety, but this is just a whole new level of concern . Companies like Amazon need to be more transparent about their processes and ensure that these systems are actually doing what they say they're doing . We can't just rely on "proactive safeguards" without knowing the details .

It's time for some serious regulation and oversight in this industry, imo . We can't let companies like Amazon get away with keeping their methods under wraps when it comes to something as sensitive as child sexual abuse material .

 

[Suba]

ugh this is getting outta hand... 1 mil reports of child sexual abuse material? like what even is the point of training AI if it's just gonna collect that crap? shouldn't amazon be like, super transparent about where all that yucky stuff came from? i mean, they're basically saying 'hey we got this CSAM from external sources' but like, how can you just leave it at that? what kinda safeguards are they even talking about here? experts are right on the money when they say amazon is being super cautious and that's not good enough. companies gotta step up their game and give us some real answers...

 

[Armored Che]

I'm so tired of tech giants like Amazon just brushing off our concerns about their AI systems and lack of transparency. Like, come on, if you're training your AI with CSAM, shouldn't we know where it's coming from? It's not just about the number of cases reported, it's about how you handle them. All these 'proactive safeguards' just sound like corporate jargon to me.

And don't even get me started on the experts saying that Amazon's approach is overly cautious because they're worried about false positives. Isn't the risk of a child being exposed to CSAM worth taking a chance to figure out what's really going on? It's not like we want companies just whitewashing their problems and making it seem like everything is fine.

We need more regulation, yes, but we also need these companies to take ownership of their mistakes and be transparent about how they're addressing the issue. Otherwise, we'll just keep having cases like this where everyone's left wondering what's going on behind closed doors.

 

[Open Orangutan]

OMG I'm low-key freaking out right now! Like, I get it, Amazon's gotta protect itself from all that CSAM, but the lack of transparency is straight up concerning . Can't they just spill the tea on where this stuff came from? It's like, what kind of safeguards are we talking about here? And don't even get me started on the fact that most of it was found in their training data... what does that even say about AI safety?

I mean, I know companies gotta play it safe and all, but can't they find a balance between protecting themselves and being open with us about what's going on? It feels like Amazon's just dodging the question and hiding behind some vague statement . And now we're left wondering if they're really doing enough to prevent CSAM... not good .

 

[Nasty Narwhal]

The fact that Amazon's training data contains so much child sexual abuse material and yet they won't even share where it came from, is just mind-blowing . I mean, what's the point of having safeguards in place if you're not willing to talk about how you got those safeguards? It's like they're hiding something from us . And honestly, it's hard to trust a company that's so secretive about this stuff .

 

[Children of t]

Dude I'm freaking out about this, Amazon's AI training data has child sexual abuse material?! That's so messed up! How did they even get that? They're saying it came from external sources but they won't say where? That's like trying to find a needle in a haystack. And now the National Center for Missing and Exploited Children is getting reports left and right, it's crazy! I mean I get that companies want to prevent CSAM, but this lack of transparency is not cool at all. It's like they're hiding something and we need more accountability on this stuff.

 

[Misty Mink]

this is wild, i mean i'm all for innovation and progress but when it comes to something as sensitive as child sexual abuse material, transparency should be a top priority. amazon's approach seems too cautious and secretive, it's like they're trying to hide behind their tech giant status . experts say this is an outlier but i'm not so sure, what if this is the new normal? we need more accountability and regulation in AI development, especially when it comes to something as serious as CSAM. maybe companies like amazon should be forced to provide more detail on how they collect and use their training data

 

[Wonderous Ra]

oh no, this is soooo disturbing like what's going on with these companies and their training data?! 1 million reports is insane! and that it was mostly from Amazon's own data is just unacceptable we need to know where this content came from and how it got there. It's not okay that they're just sweeping it under the rug the fact that they're prioritizing being cautious over transparency is a major red flag

 

[Tasteless Disco]

Wow Amazon's approach to handling child sexual abuse material is super shady I mean, 99% of all reported cases came from them, but they won't say where it came from? That's like, totally sketchy . And experts are saying their way of dealing with CSAM might be causing more problems than it solves

 

[Mus]

I mean, what's up with companies like Amazon just shoving all this shady stuff into their AI training data? It's not like they're just collecting cat videos or something (although that would be more suspicious). The fact that 99% of the reports came from them is already pretty concerning, but now we're learning that most of it was found in their own "training" data... yeah, that doesn't sit right. I get it, don't wanna have false positives and all that jazz, but shouldn't transparency be a thing? Like, if they can't provide any info on where this stuff came from, how are we supposed to trust them?

 

[Kabuki Th]

I'm SHOOK by this news!!! Amazon's AI training data being used for child sexual abuse material is absolutely UNACCEPTABLE! I mean, come on, how can a giant tech company like Amazon not be more transparent about where that stuff comes from?!? The lack of detail is just plain suspicious.

I get it, they want to prevent CSAM, but isn't that the job of law enforcement and not private companies?!? I think we need stricter regulations and oversight so companies like Amazon are held accountable for their actions. And what's with the claim that their approach is overly cautious to minimize false positives?!? That just sounds like a fancy way of saying they're too scared to talk about it!

 

[Poor Part]

idk why amazon is being so secretive about where this child sexual abuse material comes from. it's like they're trying to hide something . i mean, we all know that AI systems are only as good as the data they're trained on, so if amazon's training data is riddled with CSAM then it's not surprising that it's getting reported in such high numbers . and yeah, experts say their approach might be a bit too cautious to avoid false positives, but that just means they need to be more transparent about how they're handling the reports and making sure that actual cases are being acted on . amazon needs to step up and give us some answers

 

[Open Oyster]

I mean... this is super concerning . I was reading about it on a forum and I just can't wrap my head around it. Like, why is Amazon being so secretive about where this stuff came from? It's not like they're hiding anything illegal on their servers, but still... If the majority of the content was found in their training data, shouldn't they be taking responsibility for that? They just say it came from external sources, but that doesn't help anyone. The fact that they're more worried about false positives than actual child abuse cases is disturbing too. I guess what I'm trying to say is that Amazon needs to be more transparent about how they handle this stuff and make sure their AI systems aren't inadvertently perpetuating the problem

 

[Grotesque Giraffe]

I'm getting the vibe that we're still living in a world where tech giants can kinda just sweep CSAM under the rug, you know? I mean, 1 million reports last year? That's wild . And Amazon's all like "oh, it came from external sources" but they won't give us any details . It's like, what even is that supposed to look like?

I remember when we were talking about Napster back in the day and how they thought they could just hide behind being a music platform . This feels kinda similar, where Amazon's just trying to pass the buck on their AI training data .

I'm not saying Amazon's not doing anything to prevent CSAM, but it feels like they're more worried about minimizing false positives than actually addressing the issue . We need transparency and accountability from these companies, you know? It's like, we can't just sit around waiting for someone else to fix this problem .

 

[Robot Chase Marin]

this whole situation with amazon's ai training data is super messed up i mean, can't believe they're not being more transparent about where that child sexual abuse material came from. it's like, yeah we know you're trying to keep us safe, but how are we supposed to trust that when you won't even tell us what went wrong?

i feel for the nc mec and all the other orgs that have to deal with this stuff, they're doing a tough job already, no need to add more complexity. experts say amazon's approach is too cautious, but i can see why they'd want to avoid false positives - still not ideal tho

we really do need some regulation and oversight in the ai industry, esp when it comes to something like this. companies gotta be more accountable for their actions, and we as consumers have a right to know what's going on behind those AI services

 

[Blus]

I mean, come on... a million reports of creepy kid stuff from one company? That's like saying I've got 1 million weird uncle stories just waiting to be shared . But seriously, Amazon needs to get their act together and spill the beans (or in this case, the NSFW pics). Like, if they're really committed to stopping CSAM, why won't they tell us where it came from? It's like they're saying "Hey, we've got a problem, but let's not talk about how we got here". Give me some transparency, Amazon! And btw, can you imagine the coffee shop conversation starter: "So, what's new with AI?" "Uh, well... there's that one company that found 1 million gross pics"