Re: Canvas fingerprinting

On Jul 25, 2014, at 11:26 AM, Craig Spiezle <craigs@otalliance.org> wrote:

> From a privacy perspective, I was under the impression that such data collectively used exclusively for fraud detection, security and forensics would be a permitted use.  Can you clarify the open issues?

The W3C compliance specification (TCS [1]) allows parties to track across different
services for fraud prevention, though companies are still required to collect only the
data that is minimally necessary.  However, TCS is only designed to be a baseline,
and even then servers are not compelled to signal compliance with TCS when they
receive a DNT:1 header; they may signal compliance with a different regime, or
signal that they don’t care about a user’s request at all.

A sever may wish to point to a more narrow compliance regime, and a user agent
may require a more robust compliance regime if it so desires.  EFF’s Privacy
Badger [2] add-on requires that companies commit to a compliance policy [3]
that I think would preclude the use of cookies or canvas fingerprinting even for
security or fraud prevention.  Third-party servers that don’t publicly commit to
honor this policy are blocked from communicating with the user’s browser.

[1] http://www.w3.org/2011/tracking-protection/drafts/tracking-compliance.html
[2] https://www.eff.org/privacybadger
[3] https://www.eff.org/dnt-policy

> 
> 
> 
> -----Original Message-----
> From: Justin Brookman [mailto:jbrookman@cdt.org] 
> Sent: Friday, July 25, 2014 7:05 AM
> To: Georg Koppen
> Cc: public-privacy@w3.org
> Subject: Re: Canvas fingerprinting
> 
> 
> On Jul 23, 2014, at 11:22 AM, Georg Koppen <gk@torproject.org> wrote:
> 
>> Mike O'Neill:
>>> If the response to canvas and other forms of fingerprinting is an 
>>> arms-race with browsers and their extensions, the web will turned 
>>> into a war zone and be ruined for everybody.
>>> 
>>> This is why we need a meaningful DNT that people trust.
>> 
>> No, DNT will not help. See the FPDetective paper 
>> https://www.cosic.esat.kuleuven.be/publications/article-2334.pdf and 
>> there especially section 7.3.
>> 
>> Fingerprinting is more and more framed in the context of fraud 
>> detection and prevention of abuse. Thus, it is getting more and more 
>> common to ignore DNT because fingerprinting is not used (or at least 
>> it is claimed
>> so) to track users i.e. to invade their privacy. Rather, it is all 
>> about devices and end users' quality of service (that's at least the 
>> story those companies are trying to sell).
>> 
>> Georg
> 
> I think others may disagree about whether tracking for fraud prevention constitutes any privacy concern.  A DNT signal is a request to sites not to collect data about users across multiple sites - including for fraud/abuse prevention.  A server can signal back that it doesn't track at all, or that it tracks for a very limited set of (in the server's opinion) unobjectionable purposes.  Or it can signal back that it tracks for advertising or doesn't honor DNT requests at all  The user or user agent can then make a determination about whether to allow the interaction or not, to disable certain functionality for the server, or anything else it wants to do.
> 
> Justin
> 
> 
>>>> -----Original Message-----
>>>> From: Rigo Wenning [mailto:rigo@w3.org]
>>>> Sent: 21 July 2014 17:43
>>>> To: public-privacy@w3.org
>>>> Subject: Canvas fingerprinting
>>>> 
>>>> https://securehomes.esat.kuleuven.be/~gacar/persistent/index.html
>>>> 
>>>> There was a lot of discussion around canvas and whether it was the 
>>>> right choice. It may also be the right choice for browser to give 
>>>> users the option to turn all those nice new features off if they do 
>>>> not want to be spied upon. To what extend do browsers trust the 
>>>> origin? I think we are in a field with lots of shades of gray.
>>>> 
>>>> Otherwise we are left surfing the Web with Amaya if we want privacy.
>>>> Amaya knows no cookies, no javascript, no canvas. This can turn into 
>>>> an advantage..
>>>> 
>>>> --Rigo
>>> 
>>> 
>>> 
>>> 
>> 
>> 
> 
> 
> 
> 

Received on Friday, 25 July 2014 15:43:36 UTC