Telegram will disclose users' phones and IP addresses to authorities at their requests, the messengers' founder and CEO Pavel Durov said on September 23.
And if they had implented that to begin with and used servers that kept no logs he wouldn’t have had anything of value to hand over and they would have had to release him since he physically could not provide those things.
He built the damn situation for himself, and the fact that such issues weren’t considered practically screams “honeypot.”
Maybe we could say he wouldn’t be in this situation because he could’ve responded to every request his company got and they could’ve provided all of the zero logs they had.
I believe Telegram just wasn’t cooperating at all which is wild! Such a Musk thing to do.
That is a pretty weak argument. The issues are minor and in a library that people are moving off of to a better build and stronger validated library. Yes, it should have been like that in the first place, but the problem is minor and being addressed.
I would look more to the various features of Matrix that aren’t encrypted like room names, topics, reactions, … and not to mention the oodles of unencrypted metadata. I really wouldn’t call Matrix a high-privacy system.
I like Matrix and use it regularly, but it definitely doesn’t have a privacy-first mindset like Signal does. I’m hoping that this improves over time, but without a strong privacy first leadership it seems unlikely to happen.
Olm is now deprecated and all development is now focused into Vodozemac: https://github.com/matrix-org/vodozemac. That being said, is there no proven Olm Protocol alternative implementation for e2e encryption (proven technology) instead of reinventing the wheel.
vodozemac might become that proven implementation. Without reinventing the wheel there will never be an alternative, because everyone just reuses the one existing library.
There’s also SimpleX chat and Briar, but I’ve used both of those less than Matrix. They seem to be aiming to solve the last few issues that Matrix has, like usernames and metadata leakage.
I consider Matrix to be closer to an “Enterprise” solution, like what a business or government or non-profit would use for secure communications (literally both French and German governments use Matrix), while SimpleX/Briar seem much more aimed at individuals just wanting control over their personal conversations.
Just keep in mind that any service that asks for a phone number can also disclose it.
I hope what leaves the Signal client is a hash of your phone number, rather than the number itself. They might even be using salts and expensive-to-execute key derivation functions, to mitigate brute force searches (which are otherwise easy given the relatively small search space of phone numbers). But if compelled, it would be trivial for Signal to change that behavior.
I wish Telegram would just enable default E2EE. Oh well, time to switch to Signal!
And if they had implented that to begin with and used servers that kept no logs he wouldn’t have had anything of value to hand over and they would have had to release him since he physically could not provide those things.
He built the damn situation for himself, and the fact that such issues weren’t considered practically screams “honeypot.”
Maybe we could say he wouldn’t be in this situation because he could’ve responded to every request his company got and they could’ve provided all of the zero logs they had.
I believe Telegram just wasn’t cooperating at all which is wild! Such a Musk thing to do.
I also don’t trust Signal… And I won’t gonna switch a 4th time. I might as well switch to Matrix chat now.
I’m not sure how much we can trust matrix either to be honest. There’s some cryptographic flaws in their Olm Library. https://soatok.blog/2024/08/14/security-issues-in-matrixs-olm-library/
As it turns out being both secure and convenient is very difficult
That is a pretty weak argument. The issues are minor and in a library that people are moving off of to a better build and stronger validated library. Yes, it should have been like that in the first place, but the problem is minor and being addressed.
I would look more to the various features of Matrix that aren’t encrypted like room names, topics, reactions, … and not to mention the oodles of unencrypted metadata. I really wouldn’t call Matrix a high-privacy system.
I like Matrix and use it regularly, but it definitely doesn’t have a privacy-first mindset like Signal does. I’m hoping that this improves over time, but without a strong privacy first leadership it seems unlikely to happen.
Olm is now deprecated and all development is now focused into Vodozemac: https://github.com/matrix-org/vodozemac. That being said, is there no proven Olm Protocol alternative implementation for e2e encryption (proven technology) instead of reinventing the wheel.
ow interesting. TIL… Olm Protocol is a clone of Signal’s Double Ratchet.
Ow interesting… SimpleX is also using Double Ratchet… https://github.com/simplex-chat/simplex-chat?tab=readme-ov-file#privacy-and-security-technical-details-and-limitations
vodozemac might become that proven implementation. Without reinventing the wheel there will never be an alternative, because everyone just reuses the one existing library.
Spin up your own server for best results.
Then you only have to worry about minor metadata leakage.
https://matrix.melroy.org… I know…
There’s also SimpleX chat and Briar, but I’ve used both of those less than Matrix. They seem to be aiming to solve the last few issues that Matrix has, like usernames and metadata leakage.
I consider Matrix to be closer to an “Enterprise” solution, like what a business or government or non-profit would use for secure communications (literally both French and German governments use Matrix), while SimpleX/Briar seem much more aimed at individuals just wanting control over their personal conversations.
Personally I really hope that Dendrite will release a version somewhat close to v1: https://github.com/matrix-org/dendrite
The main downside of Matrix is the Synapse Python server (blurp). But Dendrite is still far for complete even years later now.
Here… SimpleX comparison table… Signal is also centralized.
Just keep in mind that any service that asks for a phone number can also disclose it.
I hope what leaves the Signal client is a hash of your phone number, rather than the number itself. They might even be using salts and expensive-to-execute key derivation functions, to mitigate brute force searches (which are otherwise easy given the relatively small search space of phone numbers). But if compelled, it would be trivial for Signal to change that behavior.
Anyone who used Telegram as a private communications channel in the first place is an idiot.