Please point to a single third party who is competent to evaluate if they can actually protect against the “most sophisticated digital threats” that has actually supported Apple’s claims.
Valid third parties include, but are not limited to, any national security agency or premier hacking organization with hundreds of members (i.e. actual “most sophisticated digital threats”) declaring it can protect against them, any individual or organization who has designed and implemented such a system in the past agreeing Apple has created such a system, or any certification body who has reliably certified such systems with low rates of false positives such as the Common Criteria.
Invalid support includes, but is not limited to, certification bodies that give Windows their highest security rating, marketing articles, individual hackers of no particular renown, and claims of it being “better” or “harder” without even being able to quantify where in a multiple order of magnitude range it lies.
That said, it's still a new feature. I'm sorry I don't have the NSA spokesperson here to say that they are going to pack it up and go home now because iPhones are unhackable, since that's the only thing you are willing to accept. To be entirely honest I am not even willing to hide my disdain for the certification you've repeatedly brought up at this point beyond it being a set of good practices. Like, the Titan M2 chip was assessed at AVA_VAN.5 it got exploited last month because it was written in a bunch of C and deployed without layout randomization or attempts at CFI. I trust the words of hackers (of particular renown, mind you!) and their analysis of how strong the mitigations actually are over some certification person just looking at the system and trying to take a guess as to how it'd hold out.
I did not previously know that the Titan M2 chip was assessed at AVA_VAN.5, but I do not see how the chip itself being certified against physical attacks is relevant to the security properties of the Security IC Embedded Software which is explicitly out of scope and is uncertified at any level. To support my claim that it is a certification of the hardware, not the software:
Here we see the certificate: https://www.tuv-nederland.nl/assets/files/cerfiticaten/2021/...
This conforms to the Security IC Platform Protection Profile with Augmentation.
Here we see the actual security target: https://www.tuv-nederland.nl/assets/files/cerfiticaten/2021/...
This is consistent with the Security IC Platform Protection Profile with no material changes.
Here we see the Security IC Platform Protection Profile definition: https://www.commoncriteriaportal.org/files/ppfiles/pp0084b_p...
On Page 7, Section 1.2.2, Statement 9, we see that Security IC Embedded Software is all software running on the chip (i.e. non-firmware). Security IC Embedded Software, which is what we would consider to be the Titan codebase that was attacked, is explicitly called out as not part of the Target of Evaluation (TOE).
On Page 22, Section 3.2, Statement 70, we see that the threats specified are physical, electrical, and hardware attacks. On Page 25, Section 3.2, Standard Threats, Statement 82-87, we further confirm that the enumerated threats are physical and related to the hardware itself, not the software.
On Page 30, Section 3.4, Assumptions, Statement 99, we see a assumption required for correct operation of the composite TOE (hardware + software) is that the Security IC Embedded Software correctly protects user data. As this is a assumption, this is not a evaluated claim and assumed to be true for the purposes of evaluation and is thus out of scope.
So, to reiterate, I do not see how a software attack on uncertified code in the Titan M2 chip proves the certification evaluates software incorrectly given that the software was out of scope of the certification in the first place and thus no claims of its quality are asserted as part of the certification that the Titan M2 chip received. That is like complaining that waterproofing standards for phones are garbage because they do not tell you how fire resistant a phone is. If anything, it supports my statement since the uncertified code was defeated.
If you don't trust anyone but yourself, you'll have to do the audit yourself. How do you suggest to do that? An auditor with a good track record seems like the most trustworthy practically feasible alternative to me.