Only the secret chat is e2e encrypted. All the other chat options are not. I think calls are also not encrypted since they appear in the normal chat history not in the e2e chat.
Obviously if your phone is compromised your e2ee chat is not safe.
Obviously if your phone is compromised your e2ee chat is not safe.
Pretty much, a lot of people think that seeing E2EE means everything is safe, which I believe gives a false sense of security. You can have your phone compromised (especially when I know your phone number, Signal I’m looking at you) or be subject to other means of attacks, exposing everything. I would rather know that this app is not secure so I don’t share anything important, while keeping secure communication to other means.
Matrix is far better in terms of security than Signal, but Matrix is far behind compared to Telegram features.
I agree it was, probably better now, but for the average person, it’s too much to “process” compared to just adding your phone number and signing up.
1. Signal is bad and insecure because registering user account requires giving a phone number. 2. Matrix is better, it fixes this by registering with emails (although emails also have zero click vulnerabilities) 3. Telegram is better than Matrix, it's more usable (even though it also requires a phone number like Signal)
So pick a lane, is requiring a phone-number a litmus-test for you or not. Is zero-click vulnerability something that needs to be addressed? How do you deal with malicious contacts or people in public groups sending zero-click links?
If you want security on the other hand but with fewer features and a smaller user base, go with Matrix. You don’t need an email, by the way; it’s optional (1).
Signal is just in the middle, lacking Telegram's features and Matrix's security, resulting in a weird abomination that I would never recommend to anyone. For a normal non-techie person, I would say go with Telegram, and if you care about security, use Matrix. Recommending Signal might give a false sense of security.
(1) https://ems-docs.element.io/books/element-support/page/creat....
The correct solution to sleeping with an axe struck on the roof above your bed isn't to not worry about it because axes coming loose on their own aren't a common occurrence. Telegram has no business in peoples' personal lives and it shouldn't be collecting that data.
Plus the risk of massive data breach is insane. I'm not sure if you know about the Finnish Vastaamo Psychotherapy hack, when thousands of patients' personal lives were published in the dark web https://en.wikipedia.org/wiki/Vastaamo_data_breach These victims are under constant extortion about that data getting spread even further. Now imagine that with close to one billion users. There is a LOT that people share on these platforms, how they unload to their close ones. Durov has no right to keep this amount of data sitting on some random server, especially given the authors' poor track record of security design.
>you should never rely on that and don’t trust any cloud-based service in general
This should be the take-away before the breach happens. But surely you agree Telegram is doing horrible job being transparent about its security, it's implying it's heavily encrypted, which laypeople assume means what end-to-end encrypted messaging provides.
>Recommending Signal might give a false sense of security.
Again, pick a lane. If you think zero click attacks of Signal are an issue but they magically disappear from Matrix clients, say so. They don't.
Decentralized system doesn't help with metadata. It's just spreading it to even more systems, every server people indiscriminately choose get a copy of groups' communication metadata, yay.
Your buddy self-hosts for you and your peers, now you have an individual with personal interest to take a peek at their peers' metadata. Not good.
There's very little a decentralized messaging platform offers other than baked-in resilience in case the company goes down. You can self-host the service.
But Signal is backed by Signal Foundation and really rich people like Brian Acton have helped it get the organization on a solid foundation. There's nothing that implies its going down.
From my PoV, I bin Element together with Signal, both provide content privacy, but no strong metadata privacy. For that you go with Cwtch, Briar, OnionShare, Ricochet Next.
Telegram is in the don't use for anything that isn't comparable to public Twitter, and since Telegram inevitably leads to misusing it, it's dangerous and bad tool. It was built to aggregate user data, and it will inevitably do that, because the masses generally don't prioritize privacy. Telegram monetizing user data is constantly one business decision away. And people using it are on borrowed time. We're not in disagreement about how it should be used, but people don't take that warning seriously, and when (not if, but when) shit hits the fan, it'll be like nothing anyone has ever seen before.
Also telegram has an additional password option if you want to login which avoids phone number hijack. Also if you hijack an account the secret chats don’t appear. They are bounded to the device.
And I think WhatsApp probably does it, otherwise why the authorities never complied that WhatsApp did not let them see the conversations?
Rule of thumb: never trust anything Facebook. I’m sure sending your messages through mail is more secure and private than WhatsApp these days.
Knowing someone's phone number doesn't automatically let you compromise their device. This is such a ridiculous argument.
>I would rather know that this app is not secure so I don’t share anything important, while keeping secure communication to other means.
This is nirvana fallacy. It's essentially saying "We should not talk about Telegram lying about its security, when in reality nothing is 100% secure". Yeah, nothing is, there's always an attack. That doesn't contribute anything of interest to the topic, it just tries to kill the criticism. And I'm saying this as someone who has worked on this exact topic for ten years: https://github.com/maqp/tfc
One way or another, phone numbers are like home addresses in the digital world. Once exposed, it’s just a matter of time and resources dedicated to that. Not to mention, sometimes it’s just needed to cross over the identity, that’s it.
> This is a nirvana fallacy. It's essentially saying
I didn’t say that. As I mentioned in the other comment to you, some or a lot of people just don’t care about security, and as long as this info is known, it should be treated just like any social media.
Great project with TFC, I never heard of it, but it looks interesting. I would definitely give it a try! I have a question though: does your project require a phone number? If not, why? And would you recommend Signal to anyone who is after security, privacy, and anonymity?
Because that's the trade-off you make when you want high entropy unique usernames to prevent enumeration attacks. They become long and random. There's still a "phone number". It just looks something like 4sci35xrhp2d45gbm3qpta7ogfedonuw2mucmc36jxemucd7fmgzj3ad. You know that string and you can make a computer somewhere in the world accept some GET requests. Who knows if Flask, or whatever is part of the stack, has zero-click vulnerabilities.
And yes obviously I would recommend Signal to anyone who wants content privacy. Since Signal offers only narrow by-policy metadata privacy (unless you're on burner hardware), I'd ask them if they wanted metadata privacy, and if so, I'd point them to the direction of Cwtch https://cwtch.im/. I wouldn't recommend TFC unless endpoint compromise was part of their threat model. It's complicated and nuanced in the deep end of the pool.
No idea how secure the encryption is, but calling someone on Telegram is safer than sending texts.
Yes, and that's where the 'practical' argument pops up. With all the E2EE buzz, is it really helping in the scenarios where it's supposed to work the best?
This thread gives an overview on why Signal and other apps are not really practical: https://x.com/Pinboard/status/1474096410383421452
> The broader problem of ephemeral or spur of the moment protest activity leaving a permanent data trail that can be forensically analyzed and target individuals many years after the fact is unsolved and poses a serious risk to dissent. But E2E is not the solution to it.
> I feel like Moxie and a lot of end-to-end encryption purists fall into the same intellectual tarpit as the cryptocurrency people, which is that it should be possible to design technical systems that require zero trust, and that the benefits of these designs are self-evident