On 3/17/17 03:51, westhawk wrote:
>
>> On 17 Mar 2017, at 02:35, Cullen Jennings <fluffy@iii.ca
>> <mailto:fluffy@iii.ca>> wrote:
>>
>>
>> The security of WebRTC is very weak without this,
>
> That is an overstatement of the situation in my view.
>
> There are several services that address the MiTM risks by adding their own
> identity validation mechanisms which in turn verify the DTLS fingerprint.
>
> One example is wire.com <http://wire.com> :
> https://medium.com/wire-news/the-road-to-a-more-private-and-secure-calling-protocol-a8f22d23f112
> Or Matrix.org <http://Matrix.org>
> Or https://tools.ietf.org/html/draft-johnston-rtcweb-zrtp-02
>
> All of these use cryptography in Javascript to validate the identity
> of a webRTC caller and detect MiTM.
> The limitation is that to work both parties need to be loading the
> same javascript, probably from the same site.
Yes, and the DTLS-SRTP mechanism already provides pretty good protection
from arbitrary-party interception of media (since the SDP contains
fingerprints already). But the issue here is undetected interception of
the media by the service itself. Your solution amounts to letting the
foxes guard the henhouse, which isn't a useful proposal. We need
independent validation of identity and stream isolation (that is:
isolation of the stream contents from access by JavaScript) for this to
be anything like trustworthy.
So basically, yeah, as Cullen said: the security of WebRTC is very weak
without this. That's a concise and accurate assessment.
/a