Submitted via IRC for TheMightyBuzzard
This new protection system works similar to the anti-child-porn detection systems in use at Facebook, and other social media giants like Google, Twitter, Instagram, and others.
It works on a database of file hashes, a cryptographic signature computed for each file.
Facebook says that once an abuser tries to upload an image marked as "revenge porn" in its database, its system will block the upload process. This will work for images shared on the main Facebook service, but also for images shared privately via Messenger, Facebook's IM app. Potential victims will need to upload nude photos of themselves
The weird thing is that in order to build a database of "revenge porn" file hashes, Facebook will rely on potential victims uploading a copy of the nude photo in advance.
This process involves the victim sending a copy of the nude photo to his own account, via Facebook Messenger. This implies uploading a copy of the nude photo on Facebook Messenger, the very same act the victim is trying to prevent.
The victim can then report the photo to Facebook, which will create a hash of the image that the social network will use to block further uploads of the same photo.
This is possible because in April this year, Facebook modified its image reporting process to take into account images showing "revenge porn" acts.
Facebook says it's not storing a copy of the photo, but only computing the file's hash and adding it to its database of revenge porn imagery.
Victims who fear that former or current partners may upload a nude photo online can pro-actively take this step to block the image from ever being uploaded on Facebook and shared among friends.
We won't be doing this. I don't even want to see hashes of you folks naked.
"Worried that an ex-boyfriend or girlfriend might post your intimate photos on the internet? Facebook says it has a solution – as long as you'll hand over the photos first.
The social media giant recently announced its new plan to combat "revenge porn," when individuals post nude photos online without the consent of the subject." http://www.foxnews.com/tech/2017/11/08/facebook-says-it-needs-your-explicit-photos-to-combat-revenge-porn.html
(Score: 2) by frojack on Friday November 10 2017, @08:07PM (4 children)
Doubt this would be effective.
A simple dimension change, cropping, color shift, or photo-shop would outsmart the hash every time.
Go to images.google.com and drag and drop any image you may happen to have on your computer or from any web site, and google will try to match it. Its pretty hopeless, and if the image is a face or body, and even if that exact image is readily available on the web.
No, you are mistaken. I've always had this sig.
(Score: 3, Interesting) by JNCF on Friday November 10 2017, @08:13PM
This is an issue whether it's done on the client or the server, it's not relevant to the quote. An AC below suggests that they'll probably divide the image so subsets of it can be hashed. I could see this helping for corners (if some, but not all, edges were cropped), or for sections around something easily identifiable like a face, but it wouldn't help with resizing, color changing, compression, etc.
(Score: 0) by Anonymous Coward on Saturday November 11 2017, @04:14AM (2 children)
> A simple dimension change, cropping, color shift, or photo-shop would outsmart the hash every time.
Play around with TinEye.com -- it does a surprisingly good job of matching images at different resolutions and different crops, even parody images that start with the same base and have something (mustache?) photoshopped in. Often by looking at the posting dates for the images it returns, it is possible to find out the origin of a particular image.
While I don't use FB, if I did it would be handy to be able to send them hashes (or whatever) of images that I didn't want to see uploaded by anyone else. I'd even be willing to download their executable and generate the hashes locally before uploading. Or maybe the hashing could be built into some simple utility like EzThumbs.exe ?
(Score: 2) by JNCF on Saturday November 11 2017, @04:21PM (1 child)
But TinEye has both images to compare at the same time, not one image and the hash of another image. I don't see how their method would apply here, unless facebook actually does retain the uploaded image.
(Score: 2) by JNCF on Saturday November 11 2017, @05:00PM
On second thought, maybe facebook could be looking for features of an image (relations between curves, etc.) that I don't fully understand, hashing that mess as individual pieces, and comparing those hashes to the hashes of the the features gleaned from the potential revenge porn. I doubt TinEye reexamines its whole database everytime it searches for an image. You might be onto something.