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

smeg@infosec.pub ⁨3⁩ ⁨weeks⁩ ago

• Grimy@lemmy.world ⁨3⁩ ⁨weeks⁩ ago

  If it knows what children looks like and knows what sex looks like, it can extrapolate. That being said, I think all photos of children should be removed from the datasets, regardless of the sexual content.

  source

  • Rooster326@programming.dev ⁨3⁩ ⁨weeks⁩ ago

    Obligatory it doesn’t “know” what anything looks like.

    source

    • Grimy@lemmy.world ⁨3⁩ ⁨weeks⁩ ago

      Thank you, I almost forgot. I was busy explaining to someone else how their phone isn’t actually smart.

      source
• stoly@lemmy.world ⁨2⁩ ⁨weeks⁩ ago

  There will be a lot of medical literature with photos of children’s bodies to demonstrate conditions, illnesses, etc.

  source

  • phoenixz@lemmy.ca ⁨2⁩ ⁨weeks⁩ ago

    Yeah, press X to doubt that AI is generating child pornography from medical literature.

    These fuckers have fed AI anything and everything to train them. They’ve stolen everything they could without repercussions, I wouldn’t be surprised if some of them fed their AIs child porn because “data is data” or something like that.

    source

    • vaultdweller013@sh.itjust.works ⁨2⁩ ⁨weeks⁩ ago

      Depending on how they scraped data they may have just let their rovers run wild. Eventually they wouldve ran into child porn, which is also yet another reason why this tech is utterly shit. If you can’t control your tech you shouldn’t have it and frankly speaking curation is a major portion of any data processing.

      source
• phx@lemmy.world ⁨2⁩ ⁨weeks⁩ ago

  They fed them on the Internet including libraries of pirated material. It’s like drinking from a fountain at a sewage plant

  source