I remember back when it was TextSecure - I tried to raise some usability and security issues. First I was ignored, then dismissed, then - a few years later - they implemented some of the changes.
I still use Signal. But the way the project is run is, dare I say, arrogant and dismissive.
If you're promoting your service to people who risk their lives and freedom by using it you need to make it 100% clear to them what their risks are. Today I still run into people who have no idea that Signal is storing their profile information and their contacts on signal's servers, and that opting out of setting a pin will not prevent that, and Signal still haven't updated their privacy policy to reflect it either (it still states "Signal is designed to never collect or store any sensitive information.")
They don't need to update their privacy policy because they never have access to the profil information.
Technically, the encrypted profile information and your messages (when they are in transit to your contacts) are being stored on their servers in the exact same way. The only difference is that messages are deleted afterwards whereas your profile is stored permanently until you decide to change it. That doesn't make the profile information any less secure, though. Yes, maybe in 20 years someone will be able to break AES-256 (or whatever symmetric encryption algorithm they use) and then the stored cyphertext version of your profile information might be valuable. Personally, I doubt it. But even if I turn out to be wrong: The possible attack vector against your profile information is the same as for messages: After all, tomorrow Signal could get convinced by an intelligence agency to permanently store all your encrypted messages from now on and then the exact same risks of AES-256 getting broken would apply.
Conclusion: When Signal says they're not collecting any sensitive information, they mean that they themselves don't have access to any such information because it gets encrypted. This is the promise of end-to-end encryption. They're not promising anything beyond that.
In particular, they can't promise that the encryption will never get broken. No one knows. And no one in their right mind would promise anything like that. But at least they do everything to mitigate that risk by openly publishing all their cryptography algorithms for peer review and actively participating in scientific research surrounding that topic.
> I still run into people who have no idea that Signal is storing their profile information and their contacts on signal's servers
The precise meaning of the phrase "Signal is storing their profile information on [their] servers" vs. what the average person will actually understand here, are two entirely different things here: Most people will think that Signal stores that profile information in cleartext on their servers – because that is the current status quo with almost all popular online platforms – when in reality this is not the case.
Normally, I would be saying at this point: Please stop spreading FUD. But I do agree with your statement that
> If you're promoting your service to people who risk their lives and freedom by using it you need to make it 100% clear to them what their risks are.
Signal could indeed do a better job here. In view of the above, however, I'm having the feeling the risks weren't really clear to you, either? (No offense)
irrelevant. Their policy states that their software is "designed to never collect or store any sensitive information." when in fact, it does. Where and how they store the sensitive information they are collecting is entirely beside the point. Assumptions about what people may or may not think that means don't really matter. As written it's a very straightforward and 100% false statement.
> The only difference is that messages are deleted afterwards whereas your profile is stored permanently until you decide to change it. That doesn't make the profile information any less secure, though.
Yes it does. Regardless of the fact that it's encrypted, it isn't as secure as you might think. See this thread for details: https://community.signalusers.org/t/proper-secure-value-secu...
> After all, tomorrow Signal could get convinced by an intelligence agency to permanently store all your encrypted messages from now on and then the exact same risks of AES-256 getting broken would apply.
Funny you should mention that because it turns out the Signal was handed a subpoena back in 2016 demanding that they hand over subscriber's names, phone numbers, and contacts. At the time, they were very proud to say they told them "Too bad, we don't have any of that data". In their own words:
"We’ve designed the Signal service to minimize the data we retain about Signal users, so the only information we can produce in response to a request like this is the date and time a user registered with Signal and the last date of a user’s connectivity to the Signal service."
"Notably, things we don’t have stored include anything about a user’s contacts (such as the contacts themselves, a hash of the contacts, any other derivative contact information), anything about a user’s groups (such as how many groups a user is in, which groups a user is in, the membership lists of a user’s groups), or any records of who a user has been communicating with."
Today, they are collecting and storing exactly that data. User's names, photos, numbers, and contacts. Yes, it's encrypted, but that's when they pull out something like this: https://community.signalusers.org/t/sgx-cacheout-sgaxe-attac... assuming they haven't forced Intel to leave a backdoor for them already or they don't want to take the time to brute force a pin.
This is why I think that your original statement that
> Today I still run into people who have no idea that Signal is storing their profile information and their contacts on signal's servers
is at least highly misleading and you're doing those people a disfavor by being similarly vague as the Signal website (albeit in the opposite way).
In any case, I agree that the statement you're quoting from their website,
> Signal is designed to never collect or store any sensitive information.
should be worded much more carefully (as should a lot more information on their website). Nevertheless, it should be noted that that particular statement is not part of their ToS / Privacy Policy and rather just an introductory statement.
> Regardless of the fact that [the profile information is] encrypted, it isn't as secure as you might think. See this thread for details: […]
Like you, I've been very concerned about Signal relying on SGX enclaves and I'm still extremely disappointed by the way they have been handling this topic. In fact, I've sent them multiple messages over the past year, asking them how come they trust SGX so much and what they've taken away from the Signal PIN UI/UX debacle. (I still think it's very poor UX to name a passphrase which should be as long as possible a "PIN".) Unfortunately, time and again they chose not to respond.
Nevertheless, the questionable security of SGX enclaves only comes into play if you choose to activate the Signal PIN feature and choose an insecure PIN. Obviously, this is still a huge red flag as the majority of users will do just that. But at least if you don't use Signal PINs you're good – in the sense that the app chooses a random lengthy passphrase for you. So yes, the encrypted profile still gets stored on their servers but, again, the attack vector is the same as in the case of messages getting stored during transmission.
Overall, you might think I'm contradicting myself – arguing both in favor and against Signal at the same time. And you would be right. Unfortunately, Signal is still by far the best tool we have for secure communication these days. (Where "best" is defined as "striking the best balance of versatility, mainstream acceptance and security".)
If the Signal team does indeed have a real problem with taking feedback and criticism, a better approach might've been to gather support and enter into long-term negotiation over Signal's future relationship with the community. This would make its development a lot smoother and prevent anger and bruised egos from building in the future.
> …If you want to discuss anything about circumvention or any other aspects of Signal in a way that is respectful to the rest of the community, please join in on the forums.
https://github.com/signalapp/Signal-TLS-Proxy/pull/15#issuec...
That does not to me seem like "shutting down the conversation completely".
(As if there is even a way to do that on the internet if you were to try!)
Why call it "the IME vulnerability" anyway? This isn't about a vulnerability, we're discussing compromised phones. "IME vulnerability" seems designed to make this sound like a Signal issue, which it isn't.
1. Bundle an Open source IME to be used when in incognito mode. 2. Warn users when they switch to incognito that their IME may still be recording the words they type.
This isn't just about compromised phones. A 3rd party keyboard doesn't have to respect the incognito flag.
https://support.signal.org/hc/en-us/articles/360055276112-In...
Sure, the app should say that too, not sure if it does.
Also, the small team of developers can only fix so many things at a time. There's ~50M more users today than a month ago, there's bound to be more work wrt. maintenance which will slow down implementing new features.
https://github.com/signalapp/Signal-Android/commit/0a29ffcf4...
Where would you say the line should be drawn?
When there's a security rake-in-a-darkened-shed that a large fraction of your users will step on, with a demonstrable risk to their life and liberty, I think reasonable people can agree that we're standing on the "hey, maybe we should at least pop a dialog about this" side of the line.
It took Moxie well over a year to come to the same conclusion, and then in a really lazy way as documented by the commit upthread. I'm starting to see him as a particularly unreasonable person.
For "pop a dialog about this", I don't know, that's an interesting idea, but it is hard to draw the borderline if you pursue this route.
For example, do you know that Tencent QQ bundles a full-blown endpoint security solution trying to "protect their users" and warn them their computing environment is compromised? To the point it installs a kernel driver to do the detection. Most of my tech-savvy Chinese friends believe this is bad, not only because the possible privacy dilemma but simply because it is not an messaging app's duty to ensure the user have a safe computing environment. Surely Signal can pop up a dialog about the IME concern, but what's next? When somebody bring up an interesting cross app side-channel leak on Android, should Signal scan the installed package list, try to flag any "suspicious app"?
Hemming and hawing whether a line warning about a vulnerability that compromises a secure messaging app for over a year for any reason, but especially because of a shitty, arguably sexist tone argument not behavior I want to see from people who make security tools.
IME keylogging is a known, serious, and frequently exploited issue that affects a substantial portion of Signal users. Signal's "Incognito Keyboard" setting didn't mention that the flag can be ignored, which was misleading and dangerous.
But yes, warning about accessibility settings if there's evidence of that being an attack vector seems like a good idea. I don't know about unsupported Android versions.
https://twitter.com/RealSexyCyborg/status/134995902394088652...
Do you happen to have a source for this? There’s lots of speculation out there, but I’ve never seen anyone claiming to have proof of this being frequently exploited.
Actually *most* Android malware use accessibility APIs to perform malicious action, random example from a quick Google search: https://medium.com/axdb/%EF%B8%8F-dissecting-defensor-a-stea... . That's simply because this is the most convenient way to perform malicious actions on Android without an exploit. Sure you have to convince the victim for permission, but with a nice lure people usually just fall for it.
It is much much more prevalent than malicious IMEs. Now help your "freedom in danger" friends by raising this up as a security vulnerability to Signal developers plz! /s
At this point you're not complaining about the end result -- they did actually implement something as a result of the feedback -- you're just complaining about the time it took them to do so. Which is IMO pretty silly, as the Signal folks make their own decisions about prioritization, and they're not at all beholden to the people who don't pay them for their service for any kind of schedule guarantees.
Moxie could have said "gosh, that sounds like a serious issue. Let us investigate it." Instead, he ignored the women reporting it, sent snarky DMs about the people involved, and stonewalled any attempt to discuss it.
The fix was made - as far as I can tell - without any engagement with the community affected by the problem.
Signal received $50m in funding a few years ago. If they're understaffed, something is awry.
Is there a good open source IME? I thought Apple/Google/Microsoft haven't been able to ship a decent one and most people use Baidu's.
> 2. Warn users when they switch to incognito that their IME may still be recording the words they type.
Is a blanket "Your phone might be compromised, we can't help you if it is." warning actually useful? This doesn't really provide the user with any actionable information.
>This isn't just about compromised phones.
This is 100% about compromised phones running malicious keyboard apps.
What activists have been saying - and you should speak to them, not me - is that a warning is better than lulling people into a false sense of security.
Again, your phone may not be compromised but your IME could still be malicious.
The fact that Moxie and his team won't even engage with the people who originally brought this up is somewhat vexing.
But in the end any such warning is meaningless as it can't possibly be acted upon.
>Again, your phone may not be compromised but your IME could still be malicious.
If you're using a malicious keyboard app I think it's fair to say that your phone is compromised.
(and obviously a bunch of the blame lies with people that do uncritically push Signal, if you are journalist not misleading your "sources" is important, but again they need to be educated too and it's not surprising that's not happening perfectly - efforts in that would also have been a reasonable response IMHO. And of course this is based on the assumption that the events have been presented somewhat accurately)
If your situation is that you need to communicate things that could get you killed or imprisoned, you should be using a burner phone that has pretty much nothing installed on it but Signal (or whatever app you choose to use for secure comms). You should also be using a third-party OS/ROM that you can be pretty sure hasn't been backdoored by a local telco or government, or a device that you've managed to import from abroad that likely doesn't have local modifications.
I would assume that most people do not do this, and yet somehow expect Signal to magically make the entire stack below it secure, which is a ridiculous expectation.
a) type nothing anywhere on your phone: send only emojis, 'gifs', and voice notes?
b) learn to read and write a language with keyboards that don't phone home; or transcribe your written language to an alphabet with a keyboard that doesn't phone home
c) buy an expensive phone with an OS supplied keyboard that doesn't phone home (assuming such phones exist?)
d) learn Android development and input method theory and build a new keyboard for yourself
Are any of these actions actually feasible for the general population?
As you say, a blanket warning that the customer's phone may be compromised is unhelpful. Warning customers who have a custom keyboard of the risks those keyboards pose (similar to the warning Android displays at custom keyboard install time)[0] could go a long way towards educating customers.
Signal markets itself as a one stop solution to privacy issues. I think it makes sense that they should outline the areas where they cannot, in fact, assure the customer's privacy.
[0]: https://support.swiftkey.com/hc/article_attachments/11501105...
>Shouldn't Signal then also warn or refuse to work on Android versions with known vulnerabilities? Or if there are apps installed on the device with the accessibility permission?
>Where would you say the line should be drawn?
Perhaps the accessibility permissions are relevant. If Signal could detect these settings and warn the customer if these settings are egregiously open, that would be a valuable feature in my opinion. To me, support for older versions of Android sound like an entirely different discussion.
Signal does include the "incognito" function[0]; Signal is already taking some steps to address the issue. However I'd argue that many people have likely forgotten that they ever installed a custom keyboard and if it was pre-installed on their phone they may not be aware of it.
[0]: https://support.signal.org/hc/en-us/articles/360055276112-In...