These companies should be criminally liable for training their models on child porn.

  • Mulligrubs@lemmy.world
    link
    fedilink
    English
    arrow-up
    3
    ·
    edit-2
    4 hours ago

    I think the demand for this is significantly higher than most people think, they’re not doing this to kill time. They’re making a fortune on it somehow.

    Eliminating it will just alleviate one symptom, not treat the disease or lessen the demand.

  • Sam_Bass@lemmy.world
    link
    fedilink
    English
    arrow-up
    1
    ·
    5 hours ago

    perverts have run under the radar for so long theyve increased their numbers and the availability of ai just feeds it

  • FlashMobOfOne@lemmy.world
    link
    fedilink
    English
    arrow-up
    7
    ·
    edit-2
    12 hours ago

    Yes, they certainly are, and most of the major platforms are being used for this and it’s widely known. Copilot’s prompt reviewers reported it to Microsoft, per leaked documents. Grok has been known to be in use for this basically since its inception. Meta’s LLM is used for it, and they know.

    I can’t speak for Gemini, Claude, or ChatGPT, but it wouldn’t surprise me at all if they’re being used for it too.

    These platforms are unregulated by anything other than civil lawsuits (though some international governments are starting to move on it), so yes, they’re going to appeal to every demographic, including the worst of the worst, in a desperate bid to one day be profitable.

    There was one guy arrested for it recently, a state-level deputy AG, but that dude’s the exception. Hopefully there will be more to come.

    I researched this topic recently and made a video about it.

    • shalafi@lemmy.world
      link
      fedilink
      English
      arrow-up
      1
      ·
      6 hours ago

      I’ve only used ChatGPT, but sexually speaking, it’s pretty locked down. Maybe you could manipulate it given some effort but I’m only using one or two prompts and I’m out.

  • Glitchvid@lemmy.world
    link
    fedilink
    English
    arrow-up
    33
    arrow-down
    1
    ·
    18 hours ago

    One concerning thing is that when these AI companies are pushed to crack down on this, they’ll retrain and bias their models to only recognize traditional “pornstar” body types as ‘adult’.

    This is a problem when they then get used as a mass content flagging and moderation tool — it’s not hard to imagine some adult woman taking nudes (as people do) and just because they don’t have a huge bust, getting auto flagged as having illegal material. Not to mention body types that aren’t even real, like the art made in furry communities.

    It’s not like there isn’t precedent, recently Discord banned a ton of people because its scanning system started throwing false positives, and that was only corrected because it made such a stink.

    • Virtvirt588@lemmy.world
      link
      fedilink
      English
      arrow-up
      4
      ·
      15 hours ago

      This is a problem when they then get used as a mass content flagging and moderation tool — it’s not hard to imagine some adult woman taking nudes (as people do).

      The problem is with sexual abuse material not nudity on its own, there is a clear distinction between those two. It seems like this is the major problem when it comes to “crack-downs” on this.

  • Snowclone@lemmy.world
    link
    fedilink
    English
    arrow-up
    16
    arrow-down
    10
    ·
    19 hours ago

    https://tsjournal.org/index.php/jots/article/view/29

    exposure to csam, through self reporting, increases the risk the viewer will seek contact with children, in case you need more arguments against csam. I’ve heard the argument that AI generated images may be a harm reduction situation. it’s not. it’s incredibly damaging to the viewer if they are seeking it out, it makes them more likely to abuse, if they are exposed to it as victims it can cause lasting damage, much of it is just remixed images of actual child abuse victims, and that revictimizes them, and causes serious damage as well. We really need to stop pretending the companies that make this software aren’t responsible for this software.

      • Snowclone@lemmy.world
        link
        fedilink
        English
        arrow-up
        2
        arrow-down
        2
        ·
        12 hours ago

        why would watching children be abused from real or stimulated sources be the same as playing a violent video game?

        • Lumisal@lemmy.world
          link
          fedilink
          English
          arrow-up
          9
          ·
          edit-2
          9 hours ago

          You’re right. A video game is far more interactive. Heck these days you can even put on a headset and really simulate the action of murdering people. Grand Theft Auto is a whole detailed simulation on muder, theft, and more. Wayyy more immersive than just plain video.

          Clearly then, by your own logic and the study’s conclusion, watching simulated sources make people more likely to do the thing after all. I mean, what sick person fantasizes about murdering strangers?

          Obviously we must ban the existence of violent video games or crime in society will spiral out of control.

          Edit: for the younger lemmings, this is sarcasm. “Fake X causes more real X” is really old propaganda, and its usually used on something most of the population already finds morally objectable + is harmful to society to advance some ulterior motive.

          In this case, probably government spying and the erosion of privacy, as usual.

          • Snowclone@lemmy.world
            link
            fedilink
            English
            arrow-up
            2
            arrow-down
            3
            ·
            edit-2
            11 hours ago

            your doing A LOT here, friend. I’m not sure I can unpack this… I’m not suggesting playing Grand Theft Auto is going to make anyone more likely to commit violent crimes. I don’t see the connection with csam. I don’t think the effects of viewing csam is equivalent or similar to playing a video game. I’ve played GTA, it’s really not an instruction manual or stimulation of detailed murder or violence, honestly, it’s a bit silly and while modern versions of the game are more detailed than the early versions of the game that were graphically more similar to Frogger than anything else, it’s still not very detailed. if you needed a murder equivalent to csam I think gore and ‘faces of death’ type of videos would be more comparable, video games are marketed to be fun and are rated and regulated pretty heavily in most areas of the world, AI generated csam isn’t being caught up in that same process of regulation or review, multiple corporations have made software that is producing AI generated csam and no one is really holding them accountable like you or me would be if we personally increased the reported rates of csam by huge orders of magnitude. we’d be in prison. the software isn’t free from the corporations that made it. they made it, they should be held accountable.

    • it_depends_man@lemmy.world
      link
      fedilink
      English
      arrow-up
      15
      arrow-down
      7
      ·
      18 hours ago

      This study explores a sample of 1,546 anonymous individuals

      I have a study of 23023012301230123050503450 anonymous individuals, of which 78% agreed that you smell like doo doo. Yep. True story. If you disagree, you hate science. How could you do this?

      I shouldn’t be this sarcastic, the journal is called " Journal of Online Trust and Safety (JOTS) " with a name like that, how could I ever be so arrogant to not believe 1546 anonymous individuals on the internet? Shame on me, really.