I don't use the qwerty layout, I use colemak. Likely this mitigates this for myself.
That's the equivalent of a shift cipher with a well known offset.
New acoustic attack steals data from keystrokes with 95% accuracy
21–30 of 239 posts
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#22Re: New acoustic attack steals data from keystrokes with 95% accuracy
#23Using an image classifier on spectrograms is pretty funny. Not a bad idea, given image classifiers are dime a dozen, but still.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#24So they generated training data from one laptop and microphone then generated test data with the exact same laptop and microphone in the same setup, possibly one person pressing the keys too. For the Zoom model they trained a new model with data gathered from Zoom. They call it a practical side channel attack but they didnt do anything to see if this approach could generalize at all
Good enough for PoC.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#25Earlier quoted context omitted.
This is just security through obscurity. For real security, you need a cryptographically rolling keyboard layout.
Brilliant suggestion. Have a TRNG or a CSPRNG (if too poor for a TRNG) choose the next layout at random for you, ideally with every keystroke. Good luck cracking that!
Even using Vim or Emacs would add some obufsCTRL[dbiobfuscation from all the spurious keystrokes.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#26I think about this attack when streamers on Twitch logs into websites etc.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#27Earlier quoted context omitted.
This is just security through obscurity. For real security, you need a cryptographically rolling keyboard layout.
Brilliant suggestion. Have a TRNG or a CSPRNG (if too poor for a TRNG) choose the next layout at random for you, ideally with every keystroke. Good luck cracking that!
Some places use touchscreen keypads for PIN entry exactly for this reason: to allow randomization, e.g. for opening a locked door, or for authorizing a transaction.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#28Earlier quoted context omitted.
This is just security through obscurity. For real security, you need a cryptographically rolling keyboard layout.
Brilliant suggestion. Have a TRNG or a CSPRNG (if too poor for a TRNG) choose the next layout at random for you, ideally with every keystroke. Good luck cracking that!
Could be done by using a device with a display - e.g. an "ereader" - to present a random keyboard layout. But, good luck being efficient typing on that. At that point, better use a different input model.
Or, use techniques such as those in the article, such as random keypresses played during the actual ones.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#29So they generated training data from one laptop and microphone then generated test data with the exact same laptop and microphone in the same setup, possibly one person pressing the keys too. For the Zoom model they trained a new model with data gathered from Zoom. They call it a practical side channel attack but they didnt do anything to see if this approach could generalize at all
I believe that is the generalisable version of the attack. You're not looking to learn the sound of arbitrary keyboards with this attack, rather you're looking to learn the sound of specific targets.
For example, a Twitch streamer enters responses into their stream-chat with a live mic. Later, the streamer enters their Twitch password. Someone employing this technique could reasonably be able to learn the audio from the first scenario, and apply the findings in the second scenario.
Re: New acoustic attack steals data from keystrokes with 95% accuracy
#30Earlier quoted context omitted.
This is just security through obscurity. For real security, you need a cryptographically rolling keyboard layout.
Why not just a keyboard that produces random noise?
Because the real data stream would still be there, just mixed with some noise. It feels harder to analyze whether the noise sufficiently obscures the real keystrokes than it does to ensure the actual keystrokes reveal no information.