1. 44

  2. 7

    Christine.. I think you have put enough data online for a sufficiently motivated and resourceful entity to, for example, produce a perfect-fidelity model of your voice. Like.. I see your teeth and tongue, and every fold of your sinuses.. Maybe your larynx is missing from this data? Hope so.

    Your face is present… Could this hypothetical entity perhaps to unlock you iPhone? Or identify you from photographs with greater fidelity than they could do the same to others?

    I am not even going to speculate about whether or not anything can be gleaned from physical shape of your brain.. nor am I going to check if any of the functional data is present, the f in fMRI.

    Personally, I would never deliberately expose such deeply detailed biometric data online… Not unless it was encrypted in such a manner that I could reasonably expect it not to be cracked open this century. In that case, I’d aim for the data to be the highest resolution I can muster.. But that’s a different sci-fi scenario than this, I worry…

    1. 15

      The thought came into my mind when I was writing this, but honestly if someone is dedicated enough to 3d print a medical image to try and unlock my devices I’m pretty much screwed no matter what I do.

      1. 12

        Lead pipes are cheap. :(

        I think your threat modeling is reasonable.

        1. 6

          I used to work with brain images at my past job. One thing we had to be careful about was preserving the privacy of our subjects. I had to sign documents saying that I was committed to upholding the privacy of the data I was working with. One thing we had to do, for example, was fuzz some of the DICOM slices around the facial area, since we were only concerned with finding lesions in grey matter.

          It’s actually not that difficult to reconstruct biometric data from the fMRI slices you have shared here. It doesn’t take that much dedication. I would like to echo caution about sharing biometric data online.

          1. 3

            The only identifiable information in those files is facial contours, which again is outside of my threat modelling because you can probably get that data from things like conventions.

          2. 2

            “but honestly if someone is dedicated enough to 3d print a medical image to try and unlock my devices I’m pretty much screwed no matter what I do.”

            The actual threat is some of it is used against you for employer, medical, psychiatric, courtroom, or other discrimination. If not now, then in the future under a different cultural view or regime. Storage is cheap and search pretty good. The fact that all of it works best when people share too much is the best reason to not do that.

            Up to you to accept or dodge any of those risks. I just like bringing it up when folks are talking abstractly or speculating about known risks where actual damage was done in various ways. I’d rather get straight to what actually happens so you can make an assessment.

            EDIT: I should’ve added I’m just focusing on the risk part of it here since that’s my thing. I did enjoy the original images you shared. Especially the wavy brain visualization. So neat.

        2. 5

          This is the type of stuff that makes Lobsters awesome. Well done Christine!

          1. 3

            So, I posted that warning message, but I forgot to say this. I agree, this is awesome! Well done @cadey!

          2. 1

            Would you mind sharing the blender file on a blender share website? This is so cool