Follow-up to last week's story:

https://lemmy.ml/post/16672524

EDIT1: Politicians expect to be be exempt.

EDIT2: Good news: Vote has been postponed due to disagreements.

you are viewing a single comment's thread
view the rest of the comments
[–] -4 points 2 years ago (2 children)

They say they the images are merely matched to pre-determined images found on the web. You're talking about a different scenario where AI detects inappropriate contents in an image.

  • source
  • parent
  • hideshow 4 child comments
  • [–] 5 points 2 years ago (1 child)

    change one pixel and suddenly it doesn'tmatch. Do the comparison based on similarity instead and now you're back to false positives

  • source
  • parent
  • hideshow 2 child comments
  • [–] -2 points 2 years ago*

    My guess was that this law was going to permit something as simple as pixel matching. Honestly I don't imagine they can codify in the law something more sophisticated. Companies don't want false positives either, at the very least due to profits.

  • source
  • parent
  • [–] 3 points 2 years ago (1 child)

    Matched using perceptual hash algorithms that have an accuracy between 20% and 40%.

  • source
  • parent
  • hideshow 2 child comments
  • [–] 2 points 2 years ago (1 child)

    Is there a source stating that they're going to require these?

  • source
  • parent
  • hideshow 2 child comments
  • [–] 4 points 2 years ago (1 child)

    Unfourtunately, I couldn't find a source stating it would be required. AFAIK it's been assumed that they would use perceptual hashes, since that's what various companies have been suggesting/presenting. Like Apple's NeuralHash, which was reverse engineered. It's also the only somewhat practical solution, since exact matches would be easily be circumvented by changing one pixel or mirroring the image.

    Patrick Breyer's page on Chat Control has a lot of general information about the EU's proposal.

  • source
  • parent
  • hideshow 2 child comments