There are stronger and weaker forms of it, though. Apple holds the keys to the identity (public key) exchange in iMessage. You're still trusting Apple not to mitm your conversations, but at least they wouldn't be able to do this retroactively. Other E2EE messaging apps at least have a way to verify each others' pubkeys, but it's not required and very few people do it.
That's cool. The article mentions a more common attack I forgot about, someone managing to add a new device to someone's account and get messages on there.
The device manufacture could silently push an update to reveal the keys? but the semantics of a label aren't important.
I think the important question is should an authority be able to read messages and if so, under what circumstances?
What if it's to prevent a major terrorist attack, should you have an absolute right to privacy so much so that, a means that would never be used against you can't exist even if it means the loss of a lot of life?
As I say, I think the question should be around where we draw the line and what protections we put in place to ensure it's not crossed.
It is not mathematically possible to selectively break E2EE. If it’s broken, it’s not E2EE.