this post was submitted on 04 Sep 2023
161 points (97.6% liked)
Technology
58303 readers
45 users here now
This is a most excellent place for technology news and articles.
Our Rules
- Follow the lemmy.world rules.
- Only tech related content.
- Be excellent to each another!
- Mod approved content bots can post up to 10 articles per day.
- Threads asking for personal tech support may be deleted.
- Politics threads may be removed.
- No memes allowed as posts, OK to post as comments.
- Only approved bots from the list below, to ask if your bot can be added please contact us.
- Check for duplicates before posting, duplicates may be removed
Approved Bots
founded 1 year ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
No you're wrong.
They are not cryptographic hashes. They are "perceptual" hashes or "fuzzy" hashes. They're basically just a low resolution copy of the original image. It's trivial for an attacker to maliciously send innocent seeming images that are a hash collision. This is, by the way, a feature not a bug. Perceptual hashes are not designed to perform a perfect match.
There are plenty of free white-papers on how perceptual hashes work, and Facebook's implementation is even open source.
Apple said they tested 100 million perfectly legal images and three had collisions with a CSAM perceptual hash. When you consider how many photos Apple was proposing to scan (hundreds of trillions of photos) that means thousands of false positives would have occurred even if nobody maliciously abused the system.
And because of all that - Apple was planning to do human reviews of every photo. They would, therefore, have seen every match (and every false positive). It couldn't have been hidden from Apple.