Forgot your password?
typodupeerror
Cellphones Encryption Privacy Security

Phil Zimmermann's 'Spy-Proof' Mobile Phone In Demand 107

Posted by Soulskill
from the protecting-against-all-but-the-dumbest-users dept.
An anonymous reader writes "BlackPhone was designed by Phil Zimmermann (inventor of PGP). The 4.7" display phone features a 2 GHz NVIDIA Tegra 4i ARM Cortex-A9 quad-core processor with 60 GPU cores, 1GB RAM and 16GB storage [more specs]. The OS is a customized version of Android called PrivatOS which offers encrypted calls, texts and emails that can't be unscrambled even by spy agencies. It also offers built-in resistance against malicious software which will be most welcomed for users worried about free Apps that are becoming increasingly invasive, if not pure data collection spyware for unknown 3rd parties. It's coming out this June, and many Fortune 50 companies have already ordered the phone to protect against industrial espionage."
This discussion has been archived. No new comments can be posted.

Phil Zimmermann's 'Spy-Proof' Mobile Phone In Demand

Comments Filter:
  • Re:open source? (Score:5, Interesting)

    by Aaden42 (198257) on Wednesday May 14, 2014 @04:55PM (#47003261) Homepage

    It doesn’t (necessarily) need to be, though it would be nice. If the Android-level interface to the baseband is sufficiently limited, and if all “secure mode” operations (encryption) are handled purely in Android and passed off as a ciphertext stream through the baseband, a subverted baseband would have limited ability to cause issues.

    Problems for an untrusted baseband are:
    1) If the OS will (or can be forced to) accept any type of control from the baseband (rather than exclusively the other way around), the baseband can take over the “secure” OS.
    2) The baseband can leak private information passed through it to a third party.

    Note that as a special case of #1, audio stream communication between baseband and OS is often implemented as some variety DMA or shared memory. Care would be required to ensure the baseband was incapable of reading or writing any portion of system memory other than what was explicitly setup by the OS for DMA. A hardware MMU or even physically separate DRAM circuitry could ensure this.

    So long as the baseband has no avenue for exerting control over the OS, the OS can’t be tainted by a subverted baseband. If all information passed through the baseband in indistinguishable from entropy, the baseband funneling it off somewhere else has limited value absent some other attack on the crypto (including $5 wrench).

    The last remaining attacks would be location leaks (which can be carried out against even an untainted baseband with CellCo assistance anyway) and the possibility of injecting forged traffic that might trick the user into doing something insecure. Well-designed UI should ensure that cryptographically authenticated communications are always distinguishable from untrusted.

    Not saying having a fully open baseband wouldn’t be a really nice thing, but there are well established and sufficiently secure ways for sandboxing an untrusted baseband within an otherwise secure design.

We want to create puppets that pull their own strings. - Ann Marion

Working...