Re: [rtcweb] Telling the user the connection is secure (Re: Resolving RTP/SDES question in Paris)

[Roman Shpount <[e-mail]>]

On Tue, Mar 20, 2012 at 6:52 AM, Harald Alvestrand <[e-mail]>wrote:

> I believe I have said this before, but...
>
> We should never tell the user the connection is secure.
> We should tell the user when we know he's exposed to risks that he usually
> isn't.
>
> Thus - we should not give any indication that we're using DTLS-SRTP with
> verified identities (if that's what we normally support). We SHOULD give a
> warning saying "hey, since the gateway you've connected to isn't doing
> normal authentication procedures, but instead insists on exchanging keys on
> the signalling channel, you are less sure who you're talking to than usual,
> and there are more boxes that might record your call in the way, but the
> script kiddie on your hotel WLAN still can't see your packets (translation:
> legacy SDES key exchange is in use, but SRTP is still on).
>
> All this will of course be iconified into a single cryptic graphic
> probably involving a padlock :-)
>
>
What we actually got here has to be iconified into padlocks of different
size and degree of openness.

In case of DTLS-SRTP with verified identities we still need to inform the
user of the indentity of the person they are communicating with. It has to
come from the browser, so that it can be presented with some degree of
trust. Since we do not trust web server or the javascript app, we should be
asking for user consent to get access to microphone and camera for each
call anyway. When asking for this consent, we should be either telling the
identity of the remote party or give a warning that remote party is unknown.
_____________
Roman Shpount