Apple was going to scan local photos, not cloud stored ones.
No matter the source, no matter the app.
And the phone will report you to the police if the algorithm marks any local photos as ones reported by the police using a "neural hash", which has a non-zero amount of hash collisions.
So, imagine your new house having cameras and microphones all over the place, that you cannot turn off, recording 24/7, but "only locally". If there's screaming, could be TV, could be just an argument, could be rough sex, drama practice, or maybe even violence and murder, it will mark those recordings and after a few repeats it'll send them to a person to look at your private recordings to see if it's just some bdsm play or if you're murdering your wife. Oh, the police officer looked at the video and it was just bdsm? Ok, continue until the next threshold.
But also this is a flawed analogy because the scanning is not 24/7, only when you are uploading to iCloud. It's more like "letting people in for dinner and them seeing blood splatter on your walls; after a few visits with different blood splatters, they might well suggest that someone have a look and check it's not just an accident-prone haemophiliac living there."
I'm an artist with unconventional religious convictions you insensitive clod!
You're missing the point though. The broadband sensors acting on another's behalf is the problem because all that'll happen is more and more liberties will be taken with the concept of ownership/post-purchase monteization, then god knows who is watching what. Hell, it's a security exploit away from becoming a home invader's wet dream.
> At the point of upload to the cloud service -where they would be scanned anyway-.
So scan them there? Why ahould the phone scan local photos? And icloud is enabled by default, guess who's going to disable it if that would've been implemented?
> No, it required a threshold of N photos to match before they were submitted for human verification.
Yay, private photos leaking to companies employees because of a flawed algorithm, makes perfect sense.
They can't scan them in the cloud because, unlike other cloud storage services, the data is encrypted before leaving the device and they don't have access to what they are storing. They still don't want to host bad stuff though so they tried to come up with a way to still scan somewhere while not making the encryption in the cloud useless for everyone.
Possibly your understanding of the motive is correct. But your understanding of iCloud security is not. Apple did not offer end to end encryption of photos until after. And it is not the default now.
To avoid doing it in the cloud? Then you can turn on end-to-end encryption on uploaded photos.
> private photos leaking to companies employees
Where N of them have matched known CSAM hashes at the point of being uploaded to iCloud, they will be presented for human verification, yes. How is this worse than the photos being scanned in iCloud and being flagged for similar verification?
You can turn on end to end encryption without scanning. Apple did. And Apple's modified key escrow ruled out end to end encryption. End to end means end to end. Not end to back door.
Known CSAM hashes is incorrect. The sources of the hashes are known to contain false positives. And true positives are not limited to depictions of sexual abuse.
Apple was going to scan local photos, not cloud stored ones.
No matter the source, no matter the app.
And the phone will report you to the police if the algorithm marks any local photos as ones reported by the police using a "neural hash", which has a non-zero amount of hash collisions.
https://github.com/AsuharietYgvar/AppleNeuralHash2ONNX/issue...