| |||||||||||
Well...Well...Posted Oct 1, 2004 23:27 UTC (Fri) by freethinker (guest, #4397)Parent article: Gaim-Encryption: Simple encryption for instant messages (NewsForge)
It's certainly more secure than a completely unencrypted channel. But I see no mention of key certification. Avoid a false sense of security: eavesdroppers will be foiled, but MITM will work fine.
(Log in to post comments)
Well... Posted Oct 1, 2004 23:28 UTC (Fri) by jwb (subscriber, #15467) [Link] Yep. The only way to verify the key is to wait until the conversation starts, then go have a look in the preferences to check the fingerprint.
It would be better if, instead of using a "PGP-like" scheme, Gaim just used PGP.
Well... Posted Oct 2, 2004 0:52 UTC (Sat) by IceB (guest, #8661) [Link] one low-tech way to foil MITM would be to ask the other person for some property of their public key, like the 8th to 12th digits on the third line, say.
Unless the MITM is being monitored in realtime by someone able to respond quickly (to substitute the fake key's property for the real one), this will catch out such an attack. Of course its whole strength is in it not being automated, which means some people will get sick of doing this after a short while, and most people won't do it to start with.
Well... Posted Oct 2, 2004 2:12 UTC (Sat) by jgreenseid (guest, #18640) [Link] my friends and i use ayttm, because it uses gpg encryption.
Well... Posted Oct 2, 2004 6:17 UTC (Sat) by tzafrir (subscriber, #11501) [Link] Also keep in mind that the key of your partner cannot be stored on the server (for AIM/ICQ/MSN/etc.) . Thus every time you move from one computer to another you have to re-verify the other users. I figure that this will present some interesting oppotunities of MITM.
Anything similar in the Jabber front? An option to sign jabber-protocol packets? (is it practical?)
One other thing: does the signed content include a clear timestamp and destination name? If not, it could be replayed.
Jabber encryption Posted Oct 2, 2004 9:56 UTC (Sat) by amdias (guest, #6441) [Link] I'm not an expert, but several jabber clients offer the possibility of using SSL to comunicate with the server (wich at least protects against password discovery). If both users use this option _and_ the communication between servers is also over SSL, doesn't this mean that one could only catch the unencrypted message locally in the servers or clients machines?
Also, Gabber provides the option of using GnuPG to encrypt communications. This is practical and safe, and I wonder why not all jabber clients do it this way...
Jabber encryption Posted Oct 2, 2004 10:26 UTC (Sat) by mjr (subscriber, #6979) [Link] The Jabber PGP extension is supported in more than one client, but of course, more would be merrier, Gaim included. Hope that they won't stop at doing their own thing but try to interoperate as well. For reference, see the Jabber Extension Proposal 27, Current Jabber OpenPGP Usage.
Jabber encryption Posted Oct 2, 2004 14:41 UTC (Sat) by robot101 (subscriber, #3479) [Link] "Gaim" are not doing their own thing - this is a 3rd party plugin. I don't know if they plan to implement Jabber's PGP encryption extension, or whether it could also be done as a plugin by a willing individual, but don't assume that the Gaim developers consider this a solution to encrypting IMs. I don't think they do. :)
This is just one approach Posted Oct 2, 2004 14:38 UTC (Sat) by robot101 (subscriber, #3479) [Link] As a Gaim contributor, I feel it necessary to point out that this is a 3rd party plugin to Gaim, and by no means the default or recommended mechanism for encrypting your IMs. The Gaim developers I've spoken to about it seem to take a dim view about the lack of authentication in Gaim-Encryption - by caching keys it ensures you're talking to the same person you first were (as with ssh), but there's not much focus on ensuring this original identification is correct. There are alternative mechanisms such as Gaim-e which allow you to use GPG encryption end to end.
Well... Posted Oct 4, 2004 14:42 UTC (Mon) by obobo (guest, #684) [Link] > It would be better if, instead of using a "PGP-like" scheme, Gaim just used PGP.
Actually, I wouldn't describe Gaim-Encryption as being particularly PGP-like, as encryption schemes go.
I wrote up the reasons why Gaim-Encryption doesn't use PGP in the FAQ. Basically, I wanted to do something that works opportunistically, and that doesn't interact with my PGP/GPG web of trust. That way I can accept keys from people over IM and have some security, while not having their keys end up "trusted" in my GPG keyring. I can understand why someone who has a large GPG keyring, and who only wants to IM people who are on it, would rather have a GPG-based based plugin, but I don't think that that description fits most users.
The other big issue is that IM is in a middle ground between one-shot communication (like Email, where PGP makes sense) and allowing typical network usage (where something like SSL makes more sense. While your email client will easily catch duplicate copies of the same letter, most IM protocols don't have such a feature. So, you need replay protection for IM, which PGP doesn't have.
On the other side of the fence, end-to-end SSL (with cached certificates) would be great, except that (most) IM is sessionless. You don't necessarily get any notification when I stop my client and restart it on a different machine, and that makes life hard for IM encryption. Add in the back-and-forth messages needed to establish the link (which can get hit by rate limiting on the IM server), and things get problematic.
-Bill
Well... Posted Oct 5, 2004 10:50 UTC (Tue) by angdraug (subscriber, #7487) [Link] I can understand why someone who has a large GPG keyring, and who only wants to IM people who are on it, would rather have a GPG-based based plugin, but I don't think that that description fits most users. The Right Thing for users who don't have a large GPG keyring is to build up one, and to hook up to a sizeable web of trust. And to encrypt all of their communications, so that the really important transactions don't stand out begging for brute-force attack. But who ever cares about doing The Right Thing, even about security?
Well... Posted Oct 2, 2004 2:41 UTC (Sat) by Nelson (subscriber, #21712) [Link] It's substantially better than nothing.. It's not trvial to perform a MITM attack against it. Technically, you are correct though. gaim-encryption does ask you if you accept the keys for people. You wouldn't have to, you could perform a key exchange via another channel.
Well... Posted Oct 2, 2004 5:47 UTC (Sat) by JoeBuck (subscriber, #2330) [Link] I'd advise against using it until people who understand this stuff try to poke holes in it (maybe ask Bruce Schneier if they did it right). There's all kinds of ways you can go wrong with the protocol to exchange keys and the like. It's not better than nothing if there's an easy crack.
Quite right Posted Oct 2, 2004 8:19 UTC (Sat) by flewellyn (subscriber, #5047) [Link] Bad encryption is not better than nothing; it's substantially worse, because it promotes a false sense of security.
Well... Posted Oct 2, 2004 14:46 UTC (Sat) by Nelson (subscriber, #21712) [Link] It's SSL. It simply doesn't check to see that the keys were signed by an authority. Other than that, it's just straight SSL.
Well... Posted Oct 2, 2004 15:12 UTC (Sat) by scripter (subscriber, #2654) [Link] Not trivial? That's simply untrue. If it uses unauthenticated SSL, or if the source code is available, it IS trivial to build a MITM tool -- I've done such things. Combine that with arp poisoning, and you're done.
|
|||||||||||