Computer Ethics, Spr 2016
Class 5
Week 5 Readings
You should read Chapter 2 of Baase, on Privacy
Debates
Is this legit?
Apple v the FBI
Here's a technical
discussion by Dan Guido of just what the FBI wants.
Basically, they want to access the user data on an iPhone 5c. The FBI has
the phone, but it's all encrypted. The key is a hash of a hardware key
embedded in a chip -- and not readable -- and the user's passcode. The
latter might simply be four digits, guessable in 10,000 tries.
The FBI can try to brute-force the encryption itself, but that's 256 bits.
Not feasible.
So here's the problem: iOS reads the passcode and forwards it to the
hardware encryption unit (the "AES engine"). So to be able to make guesses
of the shorter passcode rather than the longer encryption key, the user has
to go through the iOS interface. And that introduces a delay between wrong
guesses; this delay rapidly becomes 1 hour. (Worse, the phone may
have an iOS feature enabled that wipes the data after ten wrong guesses.)
The FBI needs a modified version of iOS that eliminates the delay, and which
allows entry of passcode guesses online rather than through the touch-screen
interface. (And which disables the wipe-on-wrong-guesses feature.)
But the iPhone will only install modifications of iOS that are signed by
Apple, so the modified version must come from them.
Apple is arguing that if they build this software for the FBI, now, then
every police organization in every country will demand access to it. The FBI
is willing to let Apple house the phone in question, so the software doesn't
leave the facility. What the FBI really wants is for Apple to sign
the update; anyone (more or less) could write it.
The iPhone firmware can also be updated, but that too requires a signature.
I'm not sure, though, where the firmware sits physically. If the FBI could
simply replace the firmware chip, they could bypass the installation
signature check. The new firmware would simply omit the signature check. But
I don't know how the hardware is laid out physically, and my guess is that
Apple has taken steps to prevent this (or to have the AES engine disable
itself if it detects this).
Sometimes hardware supports the JTAG
interface, meaning that you can attach connectors to the board to read
what's going on. Apparently much of Apple's hardware has no JTAG interface
(intentionally!).
But all this is something of a short-term problem. Newer iPhones have what
is called the secure enclave, which is a separate
non-readable CPU that processes the passcodes. The secure enclave itself
adds the hour delay between wrong guesses (after a certain number), so
there's no way around that. (One can still use a modified version
of iOS to enter the passcodes online, rather than through the touch-screen).
If the phone in question had a secure enclave, Apple might be able to say
"can't" instead of "won't". Alternatively, it is possible that Apple can
also update the secure enclave firmware, which is what Guido believes.
Note that at the beginning of his article, Guido suggests that FBI access to
the phone is ok because the owner has given permission. This is false,
generally, but irrelevant here.
Trump and Defamation
Donald Trump's attorneys have sent Ted Cruz a cease-and-desist letter for
playing in a Cruz campaign ad part of a 1999 Trump interview in which Trump
declared himself pro-choice.
This has definitely upped the ante for bizarreness.
Question: if Trump's argument is that the interview is copyrighted,
can Cruz still play it?
Start
with MGM v Grokster
Then privacy
Jerry Day
Privacy basics
Are we hiding something?
Snowden
Encryption