3 ms·
https://eprint.iacr.org/2023/469.pdf https://eprint.iacr.org/2023/469.pdf
by Intermernet 2y ago
https://eprint.iacr.org/2023/469.pdf https://eprint.iacr.org/2023/469.pdf
- vetinari 2y agoFrom your own link: > Recently, in [MV21 ] MTProto 2.0 (the current version) was proven secure in a symbolic model, but assuming ideal building blocks and abstracting away all implementation/primitive details. Translation: it is secure, except for bugs, if any.
- Intermernet 2y agoThat's a generous translation! They were shown to be double-encrypting, using nonces where they weren't required, and generally making a bunch of mistakes that would be fine if they were writing a student level implementation of a secure messenger protocol, but not one that went on to be tacitly endorsed by a bunch of nation states! It's like a clunkier version of the backdoor in Dual EC DRBG. When problems like this are found, you can either assume deliberate malice (as in the case of NIST) or accidental incompetence. Either should be immediate grounds for not using the software. This isn't Flappy Bird. This is meant to be secure comms. The "This Is Fine" mentality doesn't cut it.