It may be grayscale and resized/normalized in other ways. Only apple knows exactly what it is.
It's only a matter of time until internet trolls find a way to abuse this. The database is stored on user's devices, so someone downloading it and permuting innocuous images enough until they match the database, and then spreading them for funzies via some underhanded method (wallpaper download sites?)... is not too far fetched.
Most people don’t really want end to end on consumer backup services, because of the associated risks. If however you don’t want unsecured backups you can handle this manually.
Of course nobody wants the company to actually look at your data, but that’s a separate issue.
If people look at today's state (phone movement tracking, voice capture, cameras and face ID, etc.) from the mental model of 30 years ago, they would see neither privacy nor freedom. At all.
Sadly, it is not the case of preserving what we still have, if we want freedom and privacy we will have to reacquire them. And a lot of the price will have to be paid in blood.
I could think of a couple of companies this could be outsourced to, with strict business and privacy agreements in place, of course, and also conveniently on the lowest end of European minimum wage for the tier 1 reviewers.
>The server then uses the decryption key to decrypt the inner encryption layer and extract the NeuralHash and visual derivatives for the CSAM matches.
This "visual derivative" term shows up repeatedly. To me, the implication seems to be that Apple doesn't look at the actual suspected image before deciding whether to proceed with a report. Instead, I infer that they only verify whether (as the device reports) the image's neuralhash is indeed present in the NCMEC database. If my understanding is correct, their "manual review" process actually provides no protection at all against collisions or erroneous database entries.
Further supporting this, on page 4:
>Apple reviews each report to confirm there is a match
It only refers to a match, not about whether the image appears to be illegal.
This makes perfect sense from Apple's perspective- who would want to be in the business of reviewing reports of probably-illegal images?- but it means that the references to a manual review safeguard would seem to be false reassurance. Maybe I'm misunderstanding the paper.