[akersten]

Yeah, it's clear to see that they'd want to get in front of something like that happening again. My point though is that the pull quote I highlighted is a flimsy reassurance because "it all happens on the device" does not at all prevent "Apple knowing about it," yet the sentence is constructed in a way that (tricks?) people into thinking it would.

[altairprime]

That’s true of of every privacy guarantee made by all software. For example: the only guarantee we have that Tor Browser isn’t phoning home usage statistics is 1) that people are inspecting its outbound network traffic with a magnifying glass, and it isn’t; 2) and so over time their guarantee has been accepted as trustworthy. So, no curing assurance can ever be made to the point you’re concerned about, other than to recommend not using software if one assigns high priority to this threat model. So, then, as someone who does, it would be useful to understand your viewpoint in a more concrete/applied sense. I have a handful of questions:

How would you rewrite Apple’s copy to reflect this universal threat? Would you advise GrapheneOS to adopt similar copy (since the concern is equally applicable, what with five nines of users not self-compiling from inspected source) to chip away at Apple’s marketing here? Is your concern restricted to ‘nudity-encountered’ metrics (as in the topic of this post) or is it generic to all ‘xyz-encountered’ metrics, or to all metrics, or..?

[dns_snek]

You misunderstood and it's really simple. Implying that on-device scanning makes it impossible for them to access any information is misleading. Just drop that faulty reasoning because it creates a dangerous misunderstanding of how technology works.

To illustrate: Because I wrote this comment, the sun is going to rise again tomorrow.