Re: Vibration API privacy considerations

So I did an initial read through:

First off I think it's impressive that the spec authors thought to check if
a page is visible before allowing the vibration API to fire. Very good
thinking!

However, I was thinking (and I'd love to hear other's thoughts) that
allowing arbitrary length patterns might not be the best idea.

Any new standard that allows a developer to cause device A to send a unique
signal to device B can be used for cross device tracking.

Furthermore, even if we ignore the fingerprinting aspect there's a
usability aspect - specifically a griefing aspect. If the API allows us to
specify arbitrary patterns of arbitrary length, it'd be easy to make a very
abusive web page.

Forcing someone to repeatedly call the API means someone could easily write
a browser extension that limits the number of vibration calls. It would
also reduce the threat of using the vibrations as a signal for cross device
tracking.

Anyways, these are just my initial thoughts - happy to discuss further on
the call.



/********************************************/
Greg Norcie (norcie@cdt.org)
Staff Technologist
Center for Democracy & Technology
District of Columbia office
(p) 202-637-9800
PGP: http://norcie.com/pgp.txt



*CDT's Annual Dinner (Tech Prom) is April 6, 2016.  Don't miss out!learn
more at https://cdt.org/annual-dinner <https://cdt.org/annual-dinner>*
/*******************************************/

On Wed, Feb 17, 2016 at 12:13 PM, David Singer <singer@apple.com> wrote:

>
> > On Feb 16, 2016, at 17:40 , Greg Norcie <gnorcie@cdt.org> wrote:
> >
> > Would they be too faint? IIRC tempest attacks have picked up keystroke
> noises:
> >
> > https://www.schneier.com/blog/archives/2005/09/snooping_on_tex.html
> >
> > Couldn't a microphone also pick up vibration noises?
>
> yes, we should probably include this in the audio beaconing; I guess we
> could do some tests, but I would expect that under some circumstances a
> microphone will hear the vibration of another device.
>
> whether you could do it without the user noticing, I don’t know (whereas
> ultrasonic audio is fairly easy).
>
> but as the ‘Bump’ app showed, you can do a lot if you have synchronized
> clocks.
>
> >
> >
> >
> > /********************************************/
> > Greg Norcie (norcie@cdt.org)
> > Staff Technologist
> > Center for Democracy & Technology
> > District of Columbia office
> > (p) 202-637-9800
> > PGP: http://norcie.com/pgp.txt
> >
> > CDT's Annual Dinner (Tech Prom) is
> > April 6, 2016.  Don't miss out!
> > learn more at https://cdt.org/annual-dinner
> > /*******************************************/
> >
> > On Tue, Feb 16, 2016 at 8:06 PM, David (Standards) Singer <
> singer@apple.com> wrote:
> > yes, an obvious question is ‘beaconing’ using vibration.
> >
> > I guess this becomes more of a question for users with more than one
> device — especially a second device that has motion sensing. But the two
> devices would have to be awfully close for vibration to transfer.
> >
> >
> >
> > > On Feb 16, 2016, at 12:30 , Joseph Lorenzo Hall <joe@cdt.org> wrote:
> > >
> > > Are those two things or just one? That is, is this section claiming:
> > > 1) it is possible to fingerprint a device through the Vibration API by
> > > requesting information that could be used to uniquely identify a
> > > device by characterizing "tiny imperfections during their
> > > manufacturing"; and 2) it is possible for an external observer to
> > > identify someone close to them in physical reality ("meat space") by
> > > causing the user to visit a specific web page that then uses the
> > > Vibration API to vibrate the device (and the external observer
> > > observes this and connects a particular web session with a particular
> > > device)?
> > >
> > > Looking at the spec, it just accepts a list of integers and vibrates
> > > the device or not. So, I don't see a way to fingerprint devices using
> > > this spec by taking advantage of "tiny imperfections during their
> > > manufacturing" (of accelerometers and gyroscopes). Maybe it's in
> > > conjunction with another API that that becomes revelant? (e.g., if you
> > > were recording audio, I bet vibrating the phone with a little training
> > > could allow you to characterize the surface it's on and possibly the
> > > type of phone and if it's in a case)
> > >
> > > I think maybe drop the first fingerprinting concern (maybe I don't
> > > understand it) but keep the second concern that it allows an external
> > > observer in physical proximity to associate a device with a web
> > > session by causing the device to vibrate using the API. (A possible
> > > mitigation to allowing for highly unique vibration patterns would be
> > > to make only simple vibrations possible.)
> > >
> > > If you've read this far, know that at some point we'll probably have
> > > to deal with eavesdropping via mobile gyroscopes... so not
> > > fingerprinting but full on identification of speaker information and
> > > parsing speech:
> > >
> > > https://crypto.stanford.edu/gyrophone/files/gyromic.pdf
> > >
> > > On Tue, Feb 16, 2016 at 10:39 AM, Chaals McCathie Nevile
> > > <chaals@yandex-team.ru> wrote:
> > >> Hi,
> > >>
> > >> the Device API group are considering proposing a revision of the
> Vibration
> > >> API, and one of the things they propose adding is a section on
> Security and
> > >> Privacy.
> > >>
> > >> The current proposal is
> > >> <
> https://github.com/anssiko/vibration/commit/48489c54e0b7ed80900e0906fa79803c8fa77069
> >
> > >>
> > >> The two things identified are that vibration can be picked up with
> e.g.
> > >> motion sensors in the same device for fingerprinting, and that a
> vibrating
> > >> device can be physicall observed externally.
> > >>
> > >> Wondering if anyone has further input.
> > >>
> > >> Cheers
> > >>
> > >> --
> > >> Charles McCathie Nevile - web standards - CTO Office, Yandex
> > >> chaals@yandex-team.ru - - - Find more at http://yandex.com
> > >>
> > >
> > >
> > >
> > > --
> > > Joseph Lorenzo Hall
> > > Chief Technologist, Center for Democracy & Technology [
> https://www.cdt.org]
> > > e: joe@cdt.org, p: 202.407.8825, pgp: https://josephhall.org/gpg-key
> > > Fingerprint: 3CA2 8D7B 9F6D DBD3 4B10  1607 5F86 6987 40A9 A871
> > >
> > > CDT's annual dinner, Tech Prom, is April 6, 2016!
> https://cdt.org/annual-dinner
> > >
> >
> > David Singer
> > Manager, Software Standards, Apple Inc.
> >
> >
> >
>
> David Singer
> Manager, Software Standards, Apple Inc.
>
>

Received on Thursday, 18 February 2016 17:03:47 UTC