this post was submitted on 28 Aug 2023
1745 points (97.9% liked)
Lemmy.World Announcements
29084 readers
244 users here now
This Community is intended for posts about the Lemmy.world server by the admins.
Follow us for server news 🐘
Outages 🔥
https://status.lemmy.world/
For support with issues at Lemmy.world, go to the Lemmy.world Support community.
Support e-mail
Any support requests are best sent to [email protected] e-mail.
Report contact
- DM https://lemmy.world/u/lwreport
- Email [email protected] (PGP Supported)
Donations 💗
If you would like to make a donation to support the cost of running this platform, please do so at the following donation URLs.
If you can, please use / switch to Ko-Fi, it has the lowest fees for us
Join the team
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
Interesting. But aren’t hashes unique to a specific photo? Just a single change to the photo would inevitably change its hash.
I think Apple was going to implement a similar system and deploy to all iPhones/Macs in some iOS/macOS update. However was eventually 86’d due to privacy concerns from many people and the possible for abuse and/or false positives.
A system like this might work on a small scale though as part of moderating tools. Not sure where you would get a constantly updated database of CSAM hashes though.
Most people are lazy and stupid, so maybe hash checking is enough to catch a huge portion (probably more than 50%, maybe even 80% or 90%?) of the CSAM that doesn't bother (or know how) to do that?
I'm almost positive they've been developing an image recognition AI that will make slightly altering csam photos obsolete.
Here's hoping.
A hash would change if even one bit changed in that file. This could be from corruption, automated resizing by any photo processing tools (i.e., most sites will resize photos if you give them one too big), saving a lossy file time again (adding more jpg), etc.. This is why there aren’t many automated tools for this detection. Sites that have tried by using skin tones in a photo have failed spectacularly.
I’ve never heard of this FBI middleware. Does anyone have the link to this? I’d like to understand what tools are available to combat this as I’ve been considering starting my own instance for some time now.
In my utopia world, the FBI has a team updating the DB.
The utopia algorithim would do multiple subsets of the picture so cropping or watermarking wouldn't break the test (assume the 'crux' of the CSAM would be most likely unaltered?) , maybe handle simple image transformations (color, tint, gamma, etc.) with a formula.
What you're talking about is digital (aka forensic) watermarking.
IMO scanning images before posting them to a forum is a distinct and utterly completely different world than having your photo collection scanned. Especially in context and scale