Apple is - rightfully and understandably IMO - criticized for their plans, but does anyone know how Google handles this?
The false positive rate does not look great.
Instead of scanning you whole library they came up with a way to do it on device, which is the main difference between other services. If you don't enable iCloud photo storage the system can't work at all.
Very interesting stuff in their technical explanation: https://www.apple.com/child-safety/pdf/CSAM_Detection_Techni...
-Facebook's reporting overwhelmingly flags burner accounts signed up via tor etc, only absolute idiots would post actual CP on their real name account on facebook.
-Apples solution is highly invasive and dangerous, and your statement about "only running with iCloud upload" is false. It took less than a week for Apple to announce that they will open these APIs to 3rd party apps.
That just means they are allowing other apps to scan for CP if they want to (or if they are required to by law). As controversial as it may be, I would trust Apple's implementation way more than I would trust a random photo editor app's implementation.
Easiest thing is not to worry about that and just use the services as normal. You’d have to trigger the system multiple times before there was even a chance of having police involved and even then there’d be no actual evidence if you don’t have that content.
I have some respect for privacy absolutists that want to go down that path on principal but it sounds like a massive pain in the ass with no upside for most people.