>"Technology that detects CSAM before it is sent from a child’s device can prevent that child from being a victim of sextortion or other sexual abuse, and can help identify children who are currently being exploited,” says Erin Earp, interim vice president of public policy at the anti-sexual violence organization RAINN. “Additionally, because the minor is typically sending newly or recently created images, it is unlikely that such images would be detected by other technology, such as Photo DNA. While the vast majority of online CSAM is created by someone in the victim’s circle of trust, which may not be captured by the type of scanning mentioned, combatting the online sexual abuse and exploitation of children requires technology companies to innovate and create new tools. Scanning for CSAM before the material is sent by a child’s device is one of these such tools and can help limit the scope of the problem.”
I see so many top level comments about how they're stopping on-device scanning of any photo which is more intrusive. This is not true. It appears the opt-in is limited to receiving alerts and blocking sending harmful images, but the implications here is that Apple is moving iCloud scanning of media onto the device.
If anyone is going to vote on this comment or respond, please atleast go to your iPhone settings -> Accessibility -> Hearing -> Sound Recognition. Those are on device ML models working on an audio buffer that is ALWAYS ON. The idea is to do the same for visual media. Any images you take or potentially even hit the image buffer of your device could potentially be classified by increasingly powerful ML models.
It doesn't matter if the actual content is then end to end encrypted, because the knowledge of the content is now available to Apple. This is DANGEROUS.