Re: Content based detection out of scope (Was: What problems are we trying to solve?)

On 2006-12-12 19:11:51 -0500, Mike Beltzner wrote:

> When a computer has been infected by malware, it is possible for
> the network stack or local trust systems to become compromised.
> The Working Group will not consider these cases when making 
> recommendations, and will assume that the user agent has a
> trusted connection to the platform's networking stack.

I'm somewhat wary about being too specific here: It's not just about
the networking stack -- it's really the assumption that we do not
consider attacks on user agents that involve subverting the local
systemm.

Our model should be that the local system can be trusted to function
according to spec, but browsers do grant control over some user
interface aspects to web content that they should better not grant.

Looking into the "but" part in more detail is in scope for the
anti-spoofing techniques deliverable.

Cheers,
-- 
Thomas Roessler, W3C  <tlr@w3.org>

Received on Tuesday, 19 December 2006 11:01:57 UTC