The claim is that (a) both entities are properly encrypting the data _in transit_ and (b) either company could _steal_ the plaintext client-side (after decryption).

Trust that a third-party application isn't stealing the decrypted messages requires the same type and amount of trust that Apple is not stealing the decrypted messages (or maybe less trust if the third-party solution is open source, etc.).

Except the stakes for Apple are so much higher. If they’ve lied to everyone and are stealing messages, that’s a multi-billion dollar class action, against very little upside to Apple.

For a tiny company like Beeper, the incentives are different. The upside of being dishonest far outweighs the risks.

Not that I believe Beeper is nefarious. They probably aren’t. But their risk/reward for abusing trust is very different from Apple’s

Is implicitly trusting authority the hacker spirit?

And by implicitly trusting authority, you mean trusting the device manufacturer with billions of sales and intense scrutiny from security researchers and state actors spanning decades, right? This isn’t “it’s not happening because Apple promises it’s not”. This is one of the most scrutinized platforms in the world. Making wild claims and disavowing them immediately is lazy rhetoric, just as oversimplifying this as an appeal to authority is lazy rhetoric.

Going back to the original claim:

> Apple claims iMessage is E2EE, do we have proof they aren't siphoning the messages from the client once it's been decrypted?

The answer here is no. Yes, making a wild claim afterwards is lazy, but the fact remains: there is no system in place to get anywhere close to "proof".

The best we have is researchers reporting trust violations when they find them, escalating those violations in the media, and sometimes forcing the company to change behavior. Relying on (ever more skilled!) unpaid volunteer work to verify the claims of the largest company in the world seems like an appeal to authority. It also doesn't scale as they make more claims and build more complex software.

Yes, breaking E2EE for everyone is so large that it would be impossible to do at scale without anyone noticing. Breaking it selectively to target individuals (the threat people are actually worried about!) is much harder to detect, no?

> The answer here is no

That's because it's a ridiculous premise. We don't have any evidence that Tim Cook isn't robbing banks in his spare time either. I'm not saying he does.. I'm just throwing it out there because he might be.

Not to mention the fact that you can't prove a negative anyway.

You certainly can prove that a system is cryptographically or otherwise sound. There is an entire field of formal verification. Proving that an implementation is correct is often more difficult, but not impossible.

If it’s a ridiculous premise, then why do we even try?

Apple added Contact Key Verification to eliminate one possible class of attack involving a lack of user transparency. Still trusting a whole lot of trust in the stack, but is an improvement.

What you think of as a ridiculous premise I think of as a goal to aspire to