There’s no need to be confrontational or try to “gotcha” people here; Hacker News is for thoughtful discussion. As for your scenario: yes, Apple could do this. But I’m not sure what your solution to this would be? Some UI to show the addition of a new key? Hashes that you could match? There’s no reason they couldn’t backdoor the UI as well as the key distribution for a casual user; and a sophisticated one who’s looking for this kind of attack can just check the keys Apple sends them manually…
My intention was not to be confrontational. But such posts spreading misinformation aren't really thoughtful and shouldn't be tolerated.
The standard method to detect MITM attacks from server side is with public key fingerprints. Sure, that feature could be backdoored too, I've seen that in a real life product. But that's only half of the equation: you need FOSS client with reproducible builds to ensure the feature actually works. After that, the users can verify their E2EE is working the way it should. Fingerprints alone aren't enough.
As I point out in the long post, use Signal that allows this.
