Many of the camera companies are now making cameras that can put a hash in the photo to identify it as real. Hopefully before long we start to have a way to verify this on the client side.
This has been discussed but not implemented, yet. Adobe and other software companies would also have their own hashes. It is an interesting solution, that is for sure. Time will tell if it’s effective.
Any good photographer will shoot in raw. And in order to get a picture it has to be processed on a computer, there is no way around it. I wonder how that’s supposed to work with these watermarks.
It’ll probably be stored in something like a TPM, whose primary purpose is to make intact extraction of the keys difficult or impossible. A few keys might become compromised but in this scenario (unlike DRM decryption) it’s easy to ignore those keys. There’s always the chance an exploit becomes available and is more widely used, though, in which case it would definitely be less valuable.
I had many of the same questions. I have not investigated further. I’m sure some enterprising hacker will figure out how to hack it like they do everything else.
I’m sad to inform you that the first thing I thought when seeing this was “maybe it’s AI” and my day is ruined
Fuck
I thought exactly the same, and there’s not much that can be done to absolutely convince me that it’s not :(
This picture has been around a lot longer than modern AI photo generation
searches Tineye
Almost a decade old, at least.
Many of the camera companies are now making cameras that can put a hash in the photo to identify it as real. Hopefully before long we start to have a way to verify this on the client side.
This has been discussed but not implemented, yet. Adobe and other software companies would also have their own hashes. It is an interesting solution, that is for sure. Time will tell if it’s effective.
Hmm, that could potentially also get rid of photoshopping. Would be awesome!
Don’t most of these photographs use editing to at least touch them up a little? I don’t think many published photographs are actually the raw photos.
Any good photographer will shoot in raw. And in order to get a picture it has to be processed on a computer, there is no way around it. I wonder how that’s supposed to work with these watermarks.
I wonder if this will backfire in the way printers adding yellow dots to pages backfired.
What prevents the AI from putting a hash in the photo?
Does it get validated online so that the camera company keeps a copy of the hash on their end? (Which is also problematic.)
The hash would (hopefully) be authenticated. If you want to google it, search for “HMAC”.
This is assuming that the local key doesn’t get leaked, which is assuming a lot.
Aren’t these hardware keys unusable outside the hardware?
You’d need to somehow have the AI authenticate the image through the cameras hardware to use it.
Still possible though.
It’ll probably be stored in something like a TPM, whose primary purpose is to make intact extraction of the keys difficult or impossible. A few keys might become compromised but in this scenario (unlike DRM decryption) it’s easy to ignore those keys. There’s always the chance an exploit becomes available and is more widely used, though, in which case it would definitely be less valuable.
I had many of the same questions. I have not investigated further. I’m sure some enterprising hacker will figure out how to hack it like they do everything else.
Same here. Sad that my first thought was to check for AI weirdness
Think I remember seeing this picture a few years ago before ai image generation was really popular