The scariest part of this recent news is that TM Signal seem(ed) to be interoperable. People using TM Signal could interact with actual Signal users. How are you to know whether or not your groups have people using bastardized versions of Signal? Are things like Session interoperable with Signal?
What is the threat model where this matters? You have to trust the recipient with Signal. The only one I can think of is the case where your recipient is using a compromised fork and is unaware. In this case, talking about the tool and checking with them about what they are using is really the only countermeasure.
This hole question slounda like a screenshot!
Screenahot? Yes! Stole data now? Yup!
Signal isn’t that kind of app. It protects your data in flight, but only has minimal protections after the recipient gets the message. It’s a whole other game to protect data at the endpoint. If you can’t trust your recipients to protect data, then you shouldn’t send them data needing protection. In order to do that you need control over all levels of the device receiving the data, hardware, operating system, file system, and software. Anything else will always leave openings for data at rest at tge destination to be compromised by untrustworthy recipients.
In no way does Signal prevent conversations from being archived. For all you know, a recipient could be screenshotting all of your messages, and they could even be using the official app when doing so.
If you don’t trust your contacts, probably shouldn’t be messaging them anything sensitive.
Yes of course. Signal can archive messages and they can be restored, you can screenshot messages and you can have them backed up as part of a policy like icloud backups.
My question is more about how do you know you’re interacting with an authentic signal client, and not a bastardized one.
how do you know you’re interacting with an authentic signal client, and not a bastardized one.
I don’t think that’s the point… it does not matter. Even if it’s an authentic client, if the device (e.g. 0 day vulnerability on the OS) or the user (e.g. does not lock their phone while going to the bathroom) is compromised, your conversation is not secure.
At the moment you can’t. The only realistic way I could see that happening is if the servers would check the app’s digital signature and refuse the app from communicating with the official infrastructure if it didn’t match.
Even then, nothing stops the client from lying to the server.
That’s the point of digitally signing the app, to ensure its authenticity and integrity. TM and others wouldn’t be able to resign the modified app with the Signal Foundation signature.
EDIT: Yeah after thinking more about it it’s not a trivial problem, as you need to assume that the endpoint is inherently untrusted.
It’s actually possible in a way:
https://en.m.wikipedia.org/wiki/SafetyNet
But you necessarily need to limit the devices and operating systems that are allowed. No custom ROMs, no root access, etc.
It’s bullshit and breaks open computing as a concept.
Isn’t that just delegating trust to a third party, e.g. here Google? It’s not as if Google was somehow immune to 0 days.
Fuck Safetynet and Play Integrity.
Which would be absolutely disgusting given that Signal’s official app lacks some basic functionality!
Yeah, I use the molly fork because there’s features I like about it. I’d be sad if I couldn’t use it anymore. :(
What are the ones you’re after specifically?
Well yes, anyone can compile its own version of Signal and use it and it will work as long as there aren’t some major changes to its communication protocol
It’s their computing and their devices, not yours.
I mean… you do know someone can just take a screenshot, right?
And even if you use the Android thing that blocks screenshots, they can still take a photo with another phone.
You need to trust the other person for there to be any “privacy”.
The direct answer to your question is: verification of the security of the platform that the other party is using is outside of the scope of the Signal protocol. Anything you send to the other party can be taken off of their device. Signal only concerns itself with securing the message over the network and making it hard for an adversary with network dominance to build a social graph. It doesn’t protect from all SIGINT.
Additionally, since the server is open source and the protocol is open an publicly documented, it is completely possible to build your own Signal client and give it whatever capabilities that you’d like.
There are several open source packages available that allow you to interface with Signal without using the official Signal client:
https://github.com/AsamK/signal-cli
https://gitlab.com/signald/signald (also, https://signald.org/articles/clients/ )
Those third-party clients have some essential, basic functionality that the official ones for some reason lack. Signal-cli allows registering from desktop without any smartphone, Molly allows an arbitrary Socks proxy instead of being limited to just Signal’s own proxy solution, tying a desktop client with a link instead of scanning a QR code (thus allowing easy registration from an Android VM), and maybe most importantly for some - Notifications not relying on Google (Molly-Socket allows it to use UnifiedPush).
No. Even if they were, the are plenty of ways to capture the messages.
Definitely. Capturing the messages isn’t my concern though as much as interacting with non authentic clients
What is your worry about non authentic clients?
Poor security implementations that would become the weakest link in the security chain
Any signal conversation can be archived
I’m out of the loop what’s going on?
Some photographer took a picture of a politician in the Trump admin using a Signal clone. That signal clone allowed the user to archive chats to a third party.
Thanks, that’s…interesting. He doesn’t seem like the type to find and try obscure apps for fun
