Telegram founder claims Signal has a US government backdoor | Hacker News Reader

Telegram founder claims Signal has a US government backdoor

t.me

83 pointsmappu58 comments

>Unlike Telegram, Signal doesn’t allow researchers to make sure that their GitHub code is the same code that is used in the Signal app run on users’ iPhones.

I remember Telegram not publishing their Android source code for extended periods of time, which caused Telegram-FOSS to be significantly behind:

https://github.com/Telegram-FOSS-Team/Telegram-FOSS/issues/1...

I'm not up to date on the current state of affairs, but I hope this has improved since then.

Says the guy who had a backdoor in their super-secure e2e secret chats

https://words.filippo.io/dispatches/telegram-ecdh/

> super-secure e2e secret chats

Which aren't enabled by default so almost nobody uses them. Telegram users are often surprised to find out their messages aren't end-to-end encrypted at all.

Telegram can be used on multiple devices simultaneously. Syncing the private key for each chat across multiple devices can be an issue. That's why secret chats are only available on the device they were created on, and having encryption on by default would not make sense.

Yet whatsapp and matrix seem to be able to do that just fine, impossible they said

AFAIK, whatsapp is still dependent on your mobile being the primary source. It has "link a device" feature, which has some of the features but cannot be used as an independent client. For example if you're using the web version and want to see some old chats, you can't unless you use it from your mobile. WhatsApp's encryption keys are tied to the device, not the account. In Matrix, encryption keys are tied to the user's account, not the device, and thus it wont be an issue.

> AFAIK, whatsapp is still dependent on your mobile being the primary source. It has "link a device" feature, which has some of the features but cannot be used as an independent client.

Not anymore https://blog.whatsapp.com/one-whatsapp-account-now-across-mu...

> Not anymore

From the blog:

"if your primary device is inactive for a long period, we automatically log you out of all companion devices."

There is one primary device, and multiple linked companion devices. This is different from telegram where there is no primary device.

Also see : https://faq.whatsapp.com/653480766448040

"Not all messages and chats are synced to linked devices from your phone. WhatsApp Desktop syncs more message history than WhatsApp Web. To see or search your full history, check your phone."

Just fine is a bit of an overstatement for matrix. I think it got much better now, but the user experience for multi device key syncing on element and other clients gives room for improvement. I ended up having to reinstall because I could not figure out the mess I created myself. There wer so many keys involved that even having a PhD in computer science I could not easily grasp their meaning and function. I understand at least why key syncing ist not easy and I understand why someone would not support it. I however do not understand telegrams choice of obscure crypto as well as it does not make it easier to support this. But I guess their business model (pushing contextual advertisement into chats) relies on not having crypto.

IIRC ads are only on public channels, where e2e does not make sense

I agree, but most of these issues stem from not having enough funding, as active as matrix seems to be, they are struggling to actually fund its development.

Whatsapp being e2e is almost (but not entirely) worthless, it tells me that they are likely not training LLMs on my messages, but I have zero trust of their safety from client-side backdoors.

The security of e2e cannot be higher than the security of the clients

That's a different conversation, I doubt you can really trust telegram either.

Also don't forget that most people back their stuff to Google Drive/iCloud anyways, so it's not like it's impossible to access.

Funny, given I've heard from multiple infosec people, that they believe telegram is a honeypot.

This is pretty rich coming from the founder of Telegram, which doesn't have any End to End encryption by default, nor are group chats encrypted in any way.

Client devices have many routes of compromise available, it's very possible that the data exfilaration is occuring there rather than server side.

You'd think the backers of Matrix (largely European Governments) or the privacy focused userbase of XMPP would go beyond forklifting in the same double rachet library that Signal and ilk use, but apparently they all think it's the best available option.

On the Matrix side, we explicitly didn't forklift the same library - we wrote our own cleanroom implementation from scratch, first in C/C++ as https://gitlab.matrix.org/matrix-org/olm and then in rust as https://github.com/matrix-org/vodozemac. They have no code at all in common with libsignal, and are Apache licensed rather than copyleft, and have been independently audited, independent of whatever Signal gets up to (https://matrix.org/blog/2016/11/21/matrix-s-olm-end-to-end-e... and https://matrix.org/blog/2022/05/16/independent-public-audit-... respectively)

You have steadfastly spread the good word in spite of undue negativity directed towards the Matrix ecosystem.

Thank you for your hard work on making Matrix better for everyone and explaining why in depth!

You make the world a better place!

Good to know, and keep it up!

off the topic but any updates on eu dma integration with messenger et al?

also, an unanswered question remaining was that the integration with meta viz DMA would apply to element only or the matrix ecosystem in general like i would be able to use fluffychat or cinny for example

To be fair trying to do encrypted group chats isn't at all straightforward.

Says the guy when his messenger is not e2ee by default(meaning the same as plain text), groups not encrypted e2e at all and uses some shady encryption mechanism

that's why it is the prefered one, naked is privacy, but for you to understand, you need to not work for any of these agencies ;)

there is a reason why the source of most content from the NATO/RUS war comes from telegram

because being naked works better than "being anonymous and encrypted", protonmail users wants to interview you:

"Encrypted services Apple, Proton and Wire helped Spanish police identify activist" - https://techcrunch.com/2024/05/08/encrypted-services-apple-p...

because telegram has group chats and channels, which are also accessible cross platform, it doesn't make sense to turn on e2e by default. secret chat feature is encrypted and thus only available on the device you start it from.

What does being multi platform have to do with E2E? Matrix does the same thing yet supports E2E, whatsapp allows multi platform access while also having full E2E support.

In whatsapp, there is a primary source (your phone) and you can "link" other devices with the primary source. This is different from how telegram handles it. In telegram you can use sms login to use the client from multiple clients, within the same smartphone, or different devices. With whatsapp, you can only have one sms login device, which would be primary device. If you use sms login with another device, you wont be able to use the previous device. To overcome this, it has a feature which can allow you to link your primary client with other clients but it is janky and does not have all the features. For matrix on the other hand, the encryption keys are tied to the account.

While I agree in sentiment please don't equivocate plain text with non-e2e but encrypted. They are very different.

If it's not e2ee, they can decrypt any data at will, so plain text

Plain text means any intermediary can read the data. Encrypted (without e2ee) means only you, the service provider and the recipient can read the data. e2ee means only you and the recipient can read it.

Do we agree on that?

If so you must agree that "any intermediary" like a ISP, network provider or similar being able to read or modify the data is strictly worse than just "you, the service provider and the recipient", right?

If you want security and a paranoid like setup just use Tox. If you want a no-brainer chat app with video support, there's GNU Jami. If you want resilence on natural disasters, keep Braid somewhere.

Does GNU Jami work in the middle-east where most voip are banned, and only middle-eastern intelligence agency run apps like BotIM are allowed?

Tox might.

Tox is not secure. Jason Donnenfeld (of Wireguard fame) filed a very alarming issue about how obviously bad the handshake was implemented[1], and lead developer Andre Almeida (iphydf/noavarice) continued to downplay the severity in public, while fashioning exploits based on it inside his "Club Cyberia" Tox group to weaponize said issue.

Stay away from Tox.

1. https://github.com/TokTok/c-toxcore/issues/426

Apart from telegram not having any kind of actually used E2E Encryption, they store everything, which means the cost of running their service is crazy high compared to something like whatsapp which AFAIK only stores stuff up to 3 months when the device isn't online.

So we have an app which has access to everything you send, and has incredibly high hosting costs with no real income, you put 1 and 1 together and you can see it for yourself.

Ironically they claim cloud storage as the reason for not having E2E, yet matrix seems to be able to do just that.

I never use Signal because the moment you provide your phone numnber, there is no more annonymity.

To be fair, anonymity never was one of signals claims.

I know & trust the founders of signal personally.

Every time I've used telegram it smacks of a data collection tool for an Eastern Bloc nation.

I'll stick with what I know, thanks.

Edit: Also the article linked is just a hit piece on the current signal leadership by a guy promoting a book about "the radical left". This is just reinforcing my tendency to say "fuck telegram".

Tucker Carlson interviewed the founder of Telegram, interesting interview. Telegram is based out of UAE to avoid jurisdiction of Russian and Western intelligence agencies.

Tucker Carlson also revealed that he's had the contents of his private signal messages leaked to media (allegedly by US Intel agencies monitoring his communication).

Very hard to trust Tucker, is there any way of verifying his bullshit?

The guy is a shill to anyone that throws money his way. I need independent third party verification of anything he claims.

“Some people say” isn’t really good enough evidence to support this claim. It perhaps casts doubt but I wouldn’t consider it trustworthy either.

Want to bet that any messenger service that requires your phone number to register and id you on their network is compromised?

For those interested, watch this interview of the Telegram founder by the US right- media as it offers an interesting insight about the politics and competition behind these messenger / social networks - https://www.youtube.com/watch?v=1Ut6RouSs0w ... (And this is, by the way, the original source of the story - https://www.city-journal.org/article/signals-katherine-maher... - which also seems to be US right- media. Looks like a concerted Republican / Trump followers attempt to have their flock use Telegram now.)

How?

You don't need government backdoors when os trust isnt user-trustable and you could chain whatever into a sesame backdoor: https://eprint.iacr.org/2021/626.pdf (for the tldr crowd; this explains a fairly glaring failure mode in signal's post-compromise security when any secondary devices are involved)

Durov, bring back the wall!

Anything that operates in the US, specially telecom, is subject to American law, therefore weak to US intelligence agencies, that's how it is, since 9/11

There's no US law requiring encryption weakening.

That typically comes in the form of a national security letter.

FWIW the underlying article[1] is written by Chris Rufo, a prominent right-winger[2]:

[1] https://www.city-journal.org/article/signals-katherine-maher...

[2] e.g. https://www.newyorker.com/news/annals-of-inquiry/how-a-conse...

The random number generator in signal is fully predictable. There is no need for a back door -- when the front door has no lock.

:sigh:

Do you have a source for that?

Of course he doesn't have a source for that

That is a pretty damming accusation. Can you provide more details? This should be the sort of thing you should be hearing from a disclosure or ideally a vendor advisory - not an HN comment thread on a vaguely related article. Failures of randomness are almost always fatal to a cryptosystem.

Source: made it up

Source: Read and understood the original whitepaper. Plus verified it in the (probably fake) source code.

You can do it too. Might take a decade or two depending on your mathematical background.

Why would I need to spend two decades? You already did it. Congratulations on probably one of the biggest exposés of this decade. Ignore the haters with their "delusions of grandeur" insults. You will soon have global fame. So, when are you publishing your bombastic exposé? Or do you plan to sell it for it's minimum $1 billion value? Either way, great job!

Reply on news.ycombinator.com