Why do people assume Signal messenger isn’t spying on you? Yes, it has open source code, yes it uses end-to-end encryption. But we can’t check which code runs in the version from Google Play or the App Store. And also their APK (IPA) build process is essentially a black box, it doesn’t use GitHub Actions or some other transparent build system. I also heard from Techlore that they add a proprietary part to the apk to filter bots. The only thing I can assume is that people scanned the traffic coming from the app (Android), phone (iOS) and checked whether encryption keys were being sent to Signal or not. But it seems to me that this can be also circumvented. What do you think?
P.S. I myself use Signal to communicate with relatives and friends. Definetly not a hater.
Well outside of the general open source and E2EE stuff, there are a few more things.
They’re under a non-profit foundation and charity to which donating is tax-deducatble. That means they have to publicice their financial numbers. Selling data would generate a sudden revenue, which would draw attention.
They also regularily do external audits, both from external audit organisations as individuals. This list was made in august 2022, you can likely find a newer list somewhere. I just did a quick search for you. https://community.signalusers.org/t/overview-of-third-party-security-audits/13243
Signal also runs perfectly fine without anything Google btw. It uses PlayServices only if you have it on your phone (otherwise it just uses WebSockets), as it preserves battery life. However, it doesn’t actually send data to Google over PlayServices. Instead it sends an empty notification, which wakes the phone and is recognised by Signal as a trigger to make it connect to Signal servers to grab data directly from there. If you wish, you can check this in the code yourself. I guess you may also be able to confirm this looking at network traffic from and to your phone.
Also a note on the E2EE. Another important thing is that not only the message is encrypted, but also the metadata. Unlike most other chatapps like WhatsApp; who knows where you are, who you talk to, how often, etc. You could theoretically also check this by checking outgoing traffic if you wish.
This also means that unless they somehow secretly have a copy of your private key, there is no data for them to sell anyways. The fact that even in court they’ve didn’t have data to show, them passing many external audits without this being a point (sometimes issues are found, which is normal. If audits are always perfect I’d be more warry. But never on this point afaik), and that nothing in the code nor internet traffic points to them possibly having this, makes me not that worried about the idea that they secretly got a copy of peoples private keys.
So overal while it’s perhaps technically possible they secretly run something else on their server and build a back door to read your messages, they are many things that show they don’t, and literally nothing that would say they do. And neither does there seem to be any reason why, since they can’t sell it nor give it in court. So unless you believe they have some evil bigger plan, I don’t see the reason to doubt.
And a little note. Privacy people can be crazy, and I say that in a positive way! If you can check it, people no doubt have, and issues would’ve been found. Yet many people deep into it still vouch for it. That says something. And the less crazy people profit of this. This is similar to why many big FOSS projects are considered safe even if you didn’t check all code yourself. And before you say “but if everyone thinks like that”, realise that the craziest don’t trust other people either. While smaller projects could hide perhaps, the real big/famous projects like Signal, Linux, LibreOffice, etc would fall trough as soon as they start doing shit.
Very well composed and thorough explaination. Thank you.
💯
Great explanation, thank you for the effort put into this. Going to forward this to a few friends who were also concerned about Signal’s privacy.
Yeah, good info for basic trust.
Thank you for the detailed comment!
Well, I think that in such a case it would be possible to bypass the correct accounting of funds. Financial fraud has not been canceled. But this is more of a counterargument, unlikely.
I didn’t know about Google notifications, cool implementation!
Yes, metadata encryption is cool, absolutely!
The question is also how to check the traffic on the iPhone, if there are even no monitoring tools there.
Well, sure but it’ll be quite difficult to hide a large increase in revenue still. Large unussual transactions generally have to be flagged by banks, so receiving and moving around revenue of sold data from your non-profit wouldn’t be thát easy unless they only allow crypto or cash. Surely it’s possible, but financial fraud on that level is quite difficult and often falls trough sooner or later. Or, the other option is that they don’t earn that much from it making it easy to hide, but that sounds like a lot of effort and potential risk for little gain.
Either way, the financial numbers is just one of the reasons. But trust is never build on one thing, it’s built on the combination of them. With all things I mentioned, I don’t exactly get the feeling it’s all hanging on finacial fraud.
Use a network you controll (like your home WiFi) and check in- and outgoing traffic network wide instead of on-device.
You cannot check other peoples stuff all the time, but I’d suggest not sending sensitive information to people you don’t trust as they could leak it (be it on purpose or not). And depending on level of sensitivity, just speak face-to-face in a private place. There is always a form of digital footprint when doing stuff digital. In the end, you should always assume that nothing is 100% safe, and anything cán be hacked. Trusting digital communication to be 100% safe is foolish. Look at situations like the Encrochat debacle for example. The question is more, which risks are worth it in your threat model. For most people, Signal is good enough as the risks it does have aren’t in their threat model at all.
Thank you! Really good points ❤️