5 ms·

Teens can proactively block their nude images from Instagram, OnlyFans

HNTW 4y ago
The last time this was proposed there was quite a bit of concern about how uploads weren't viewable by anyone and could/would be abused by 3rd parties as an anonymous way to take down rival content without any recourse against another OnlyFans producer, to cause headache for a FB friend, etc. It looks like this system, as opposed to NCMEC, is opt in so maybe some of these issues have been resolved?
- londons_explore 4y ago
  It seems you could solve this by allowing anonymous & privacy preserving blocking of images only if that particular image is nowhere on the internet already. If the image is already published, then a review process is necessary to verify the identity of the person requesting removal.
  - cyanydeez 4y ago
    That just seems impossible to verify. This system will just be abused until it's taken down. You really just need actual enforcement of clear criminal cases to seriously manage the problem.
_manifold 4y ago
>To use Take It Down, anyone—minors, parents, concerned parties, or adults concerned about their own underage images being posted online—can anonymously access the platform on NCMEC’s site. Take It Down will then generate a hash that represents images or videos reported by users as sexualizing minors, including images with nudity, partial nudity, or sexualized poses. From there, any online platform that has partnered with the initiative will automatically block uploads or remove content matching that hash. This sounds impressive if you don't know how file hashing works. If a malicious actor wants to get around this, all they would have to do is change a single pixel and/or re-export as a different format.
- mikestew 4y ago
  all they would have to do is change a single pixel and/or re-export as a different format This sounds plausible if you don't know how perceptual hashing works: https://en.wikipedia.org/wiki/Perceptual_hashing https://en.wikipedia.org/wiki/Perceptual_hashing
  - mrguyorama 4y ago
    Your own links talks about how perceptual hashing hasn't been proven to be robust enough for this use case, and also introduces a new problem: Hash collisions, such that you can generate images that hash to the same perceptual hash as an illicit image.
    - mikestew 4y ago
      Robust or not, it's a little more nuanced than a file hash that goes out the window because a single pixel changed, as implied by GP.
  - _manifold 4y ago
    Yes, I did some more research on this as well. Apparently perceptual hashing is also susceptible to some pretty simple attacks: https://arxiv.org/pdf/2111.06628.pdf https://arxiv.org/pdf/2111.06628.pdf
- AdamJacobMuller 4y ago
  Not necessarily. There are much newer technologies than simple hashes of files now, which are effectively content-aware image hashing algorithms which are highly resistant to manipulation techniques (re-encoding, resizing, even things like rotation/blur) they are of course tunable algorithms which the more you want to catch the more false positive rate there is but you can already today do much better than simple file hash. Look at https://www.microsoft.com/en-us/photodna https://www.microsoft.com/en-us/photodna and https://openbase.com/python/ImageHash/documentation https://openbase.com/python/ImageHash/documentation
  - 4y ago
chrismeller 4y ago
Only semi related, but can someone explain to me how the hashing process works? Of course I get the process of hashing passwords and such, but how do you hash a photo so that the hash isn’t invalidated as soon as I convert it to a different format or trim a single pixel off of it or add a scrolling banner that “this video was uploaded to…”?
- e12e 4y ago
  https://en.wikipedia.org/wiki/Perceptual_hashing https://en.wikipedia.org/wiki/Perceptual_hashing
kevincox 4y ago
> Also, please make sure the file names for your photos or videos don’t include your name or other information that could identify you. But why? Is there any need to log the file name?
fear_and_coffee 4y ago
[dead]

HNTW 4y ago

The last time this was proposed there was quite a bit of concern about how uploads weren't viewable by anyone and could/would be abused by 3rd parties as an anonymous way to take down rival content without any recourse against another OnlyFans producer, to cause headache for a FB friend, etc. It looks like this system, as opposed to NCMEC, is opt in so maybe some of these issues have been resolved?

londons_explore 4y ago
It seems you could solve this by allowing anonymous & privacy preserving blocking of images only if that particular image is nowhere on the internet already. If the image is already published, then a review process is necessary to verify the identity of the person requesting removal.
- cyanydeez 4y ago
  That just seems impossible to verify. This system will just be abused until it's taken down. You really just need actual enforcement of clear criminal cases to seriously manage the problem.

_manifold 4y ago

>To use Take It Down, anyone—minors, parents, concerned parties, or adults concerned about their own underage images being posted online—can anonymously access the platform on NCMEC’s site. Take It Down will then generate a hash that represents images or videos reported by users as sexualizing minors, including images with nudity, partial nudity, or sexualized poses. From there, any online platform that has partnered with the initiative will automatically block uploads or remove content matching that hash. This sounds impressive if you don't know how file hashing works. If a malicious actor wants to get around this, all they would have to do is change a single pixel and/or re-export as a different format.

mikestew 4y ago
all they would have to do is change a single pixel and/or re-export as a different format This sounds plausible if you don't know how perceptual hashing works: https://en.wikipedia.org/wiki/Perceptual_hashing https://en.wikipedia.org/wiki/Perceptual_hashing
- mrguyorama 4y ago
  Your own links talks about how perceptual hashing hasn't been proven to be robust enough for this use case, and also introduces a new problem: Hash collisions, such that you can generate images that hash to the same perceptual hash as an illicit image.
  - mikestew 4y ago
    Robust or not, it's a little more nuanced than a file hash that goes out the window because a single pixel changed, as implied by GP.
- _manifold 4y ago
  Yes, I did some more research on this as well. Apparently perceptual hashing is also susceptible to some pretty simple attacks: https://arxiv.org/pdf/2111.06628.pdf https://arxiv.org/pdf/2111.06628.pdf
AdamJacobMuller 4y ago
Not necessarily. There are much newer technologies than simple hashes of files now, which are effectively content-aware image hashing algorithms which are highly resistant to manipulation techniques (re-encoding, resizing, even things like rotation/blur) they are of course tunable algorithms which the more you want to catch the more false positive rate there is but you can already today do much better than simple file hash. Look at https://www.microsoft.com/en-us/photodna https://www.microsoft.com/en-us/photodna and https://openbase.com/python/ImageHash/documentation https://openbase.com/python/ImageHash/documentation
- 4y ago

chrismeller 4y ago

Only semi related, but can someone explain to me how the hashing process works? Of course I get the process of hashing passwords and such, but how do you hash a photo so that the hash isn’t invalidated as soon as I convert it to a different format or trim a single pixel off of it or add a scrolling banner that “this video was uploaded to…”?

e12e 4y ago
https://en.wikipedia.org/wiki/Perceptual_hashing https://en.wikipedia.org/wiki/Perceptual_hashing

kevincox 4y ago

> Also, please make sure the file names for your photos or videos don’t include your name or other information that could identify you. But why? Is there any need to log the file name?

fear_and_coffee 4y ago

[dead]