Slashdot Mirror

← Back to Stories (view on slashdot.org)

Researchers Use Smartwatch To Spy What Users Are Typing On a Keyboard

Posted by samzenpus on from the lets-see-what-you're-writing dept.

An anonymous reader writes: Researchers have created an app that follows the micro-movements of your smartwatch and is able to detect what keys you're pressing with your left hand, and guess what words you may be typing on a keyboard. The app developed for the Motion Leaks (MoLe) project only works on a Samsung Gear Live smartwatch, but researchers say that in theory, a similar app could be developed for other smartwatch makes and models.

7 of 38 comments (clear)

  1. eagle system by nospam007 · · Score: 4, Funny

    I doubt that. I'm using the eagle system, circling over the keyboard until I detect the prey (right key) and then I let it fall down on it (or near it) naturally always with the same finger.
    It should be easier with touch typists.

  2. One Handed Typing Activies by Canth7 · · Score: 4, Insightful

    Time to add taking off that watch in addition to firing up VPN and private browsing before visiting your favorite adult sites now.

  3. Acoustic is better by Dan+East · · Score: 5, Interesting

    This is an interesting although convoluted method to determine what is being typed. It has already been demonstrated that the acoustics from typing can be used to identify what is being typed. Most smart watches have microphones. It makes more sense to use the microphone right next to the keyboard to capture very high quality audio so close to the source and then analyze it acoustically to determine what was typed (which captures data from BOTH hands). It will also work if the user takes off their watch and lays it nearby.

    http://www.cs.berkeley.edu/~ty...

    --
    Better known as 318230.
    1. Re:Acoustic is better by Dan+East · · Score: 2

      Obviously you didn't read the research paper I referenced.

      In this paper we argue that a labeled training sample requirement is unnecessary
      for an attacker. This implies keyboard emanation attacks are more serious than
      previous work suggests.

      We built a prototype that can bootstrap the recognizer from about 10 minutes of
      English text typing, using about 30 minutes of computation on a desktop computer
      with a Pentium IV 3.0G CPU and 1GB of memory. After the bootstrap step, it
      could recognize language-independent keystrokes in real time, including random
      keystrokes occurring in passwords, with an accuracy rate of about 90%.

      Of course sound can be recorded and thus post-processed whenever a large enough corpus has been captured. Thus you could type for small 1 minute periods at a time over the course of days, and once 10 minutes worth of recordings had been captured from that specific keyboard, it could be analyzed, and after that interpreted in real-time if desired.

      --
      Better known as 318230.
  4. Fooled ya! by AndyKron · · Score: 2

    I wear my watch on my RIGHT hand. Ha Ha!

    1. Re:Fooled ya! by I'm+New+Around+Here · · Score: 4, Funny

      Fool! You think these scientists haven't thought of that? Of course they have.

      The interesting part is that it still captures what the left hand is doing. Some weird entanglement issue, I believe.

      --
      If you think I voted for Trump because of this post, you're wrong. I voted for Dr. Jill Stein of the Green Party. Again.
  5. Don't dig too deep, you won't like what you see by U2xhc2hkb3QgU3Vja3M · · Score: 5, Funny

    Researchers have created an app that follows the micro-movements of your smartwatch and is able to detect what keys you're pressing with your left hand...

    I think they'll detect other kinds of movements made with the left hand.