As far as I'm aware, the most proper attacks here are, in order of cost: 0) Find some errata. Apple presumably knows as much as anyone except NSA. Have plaus

As far as I'm aware, the most proper attacks here are, in order of cost: 0) Find some errata. Apple presumably knows as much as anyone except NSA. Have plausible deniability/parallel construction. 1) OS level issues, glitching, etc. if the device is powered on (likely not the case). Power stuff seems like a particularly profitable attack on these devices. 2) Get Apple, using their special Apple key, to run a special ramdisk to run "decrypt" without the "10 tries" limit. Still limited by the ~80ms compute time in hardware for each try. (vs. an iPhone 5S/6/6S with the Secure Enclave:) 3) Using fairly standard hardware QA/test things (at least chip-level shops; there are tens/hundreds in the world who can do this), extract the hardware key. Run a massively parallel cluster to brute force a bunch of passphrases and this hw key, in parallel. I'd bet the jihadizen is using a shortish weak passphrase, but we can do 8-10 character passphrases, too. They may have info about his other passphrases from other sources which could be useful. While I'm morally against the existence of #3, I'm enough of a horrible person, as well as interested in the technical challenge of #3, that I'd be willing to do it for $25mm, as long as I got to do it openly and retained ownership. In secret one-off, $100mm. I'd then spend most of the profits on building a system which I couldn't break in this way.