Trust, then verify. No ability to verify? No trust.
I am very sympathetic to the idea that more components should be open source, and Apple's systems should be much more open (particularly backup.) But at the end of the day if Apple is compromised there is no open source solution that can save you. They design the silicon.
If apple cared about providing encryption to the masses, ADP would be enabled by default and you'd have to opt out of it.
As-is, all your messages, photos, and so on are backed up unencrypted to apple's servers where they can read them at will. End-to-end encryption is opt-in, and I doubt most "the masses" even know a setting for that exists.
Apple is also a company that needs to cater to its customers. If they enabled ADP by default and customer locks themselves out and goes to Apple, they want to be able to help. ADP is intended for people who understand what it is but nit savvy enough to run their own system.
Internal metrics for support teams are almost entirely customer satisfaction focused, which built a culture of getting a result for the customer at all costs, which was very exploitable by social engineering.
It doesnt surprise me that they dont want to let customers encrypt and lose all their baby photos by default.
Providing encryption to the masses would in fact be telling people who lost their phone, or forgot their password "no, all your photos are gone forever, tough luck. Also, you have to make a new apple account and re-purchase all your apps".
They also switched a few years back to provide signed firmwares rather than encrypted firmwares to ease independent verification, and have the Apple Security Research Device program to do runtime exploration with certain security walls turned off. (Supposedly creating these devices requires a partial factory retooling)
Apple also only has per-device global builds, rather than regional builds which might obscure requested features. My understanding is that they take transparency measures to make sure it can be detected if a firmware was released out-of-stream, and anonymity measures to prevent targeting a specific device with a custom firmware.
The Secure Enclave also requires the device passcode as part of an approval process for installing new device/enclave firmware; the underlying OS and security enclave are not meant to have the capability of being transparently updated/modified.
However there is "Deterrence through Accountability. We can attempt to legally prosecute the attackers" [1].
That is what is happening here. The attackers are being prosecuted.
[1] https://www.cs.cornell.edu/courses/cs5430/2017sp/l/03-princi...
Australia introduced the concept of these laws with the Access and Assistance bill. The politicians were adamant it was necessary, however there were so many potential users of the system (Politicians, police, spy agencies) it never made a lot of sense.
Fast forward a few years and ASIO gave a press conference where they admitted to only having used the powers under the bill twice. Which makes me concerned about who the bill was for and what it has been used for. Unlike the British version, any public information release leads to instant jail time, and it was unclear whether this extended to briefing legal counsel.
I feel like, if the 5 eyes wanted to breach iCloud they would use Australia rather than Britain where it can be publicly contested like this.
And yes, we shouldn't put trust in corporations in the first place.
Otherwise before "Trust but verify" became popular, "Put no faith in words" was apparently the go to for Russian leaders. Not as catchy indeed.