Not pronouncing about what path is the most distopic, just for the fun of the exercise of what if we push in the direction:

Given the rule, I would expect (IANAL), Apple should not deal with data stored on phones they sold.

People are responsible for what they store on their device. When I take a photo in the street, if someone come to me asking to erase a photo with them or their kids as they were in the background, I'll tell I don't publish any photo online, which is generally what people are thinking of as a concern and that stop there, but if they insist I will remove it from my phone. Because I'm too lazy to actually live edit the photo and remove them from the picture, even if that is certainly doable with a simple prompt by now.

Now if Apple store automatically photo in some remote server they own, they are the ones who should be responsible to comply with making sure they won't store something illegally. Microsoft, Google, and Apple use PhotoDNA to detect known CSAM if I'm not mistaken. Though legally they only should remove once they get a notice about it. Same way, they could proactively blur visages of people not detected as the people that were whitelisted for the uploading account. And, by that logic, they should certainly remove the information regarding a person if they get a notice, just as well as they wouldn't keep CSAM data once notified, would they?

Anyway the underlying issue is not who store what, but what societies lose at letting mass surveillance infrastructures being deployed, no matter how the ownership/responsibility dilution game is played on top of it.