• Pierre-Yves LapersonneOP
    link
    fedilink
    arrow-up
    5
    arrow-down
    8
    ·
    1 year ago

    Olvid is the only messaging application today certified by the ANSSI (the French security agency), I doubt this one will be hacked.

    • FutileRecipe@lemmy.world
      link
      fedilink
      arrow-up
      17
      ·
      1 year ago

      Olvid is the only messaging application today certified by the ANSSI

      Is there a list of those they’ve tested, and why they didn’t meet criteria? Has Signal been tested?

      Without that info, just seems coincidental that the French government has bestowed this cert on only a French app.

      • Pierre-Yves LapersonneOP
        link
        fedilink
        arrow-up
        5
        arrow-down
        8
        ·
        1 year ago

        A matter of time, resources, knowledge, tools, energy, intels. With the ANSSI credit, good luck.

        • Nia@lemmy.ml
          link
          fedilink
          arrow-up
          4
          ·
          1 year ago

          Just because everything checks out in principle doesn’t mean it’s actually secure. First off, we have no certainty of the client code running; it’s open source, sure, but unless they ensure reproducible builds - which, given it’s on the Play store (and I assume Apple app store), they can’t be, since the binaries must be signed - we have no way of knowing whether the code actually being downloaded and run is actually the same as the FOSS version. Further, even if it is, it may have intentional subtle vulnerabilities meant to be used by the French govt (so would easily pass certification by having the ANSSI be instructed top-down to overlook certain things), or it may be that the server can trigger a known bug resulting in leakage of data. At an even more paranoid level, it’s possible that the encryption itself is faulty; the specification says it uses aes256 and ed25519 which is about as battle-tested as it gets, but the PRNG seems to be mostly their own innovation. It specifies a minimum of 32 bytes of entropy, which (though cryptography is not my expertise, so at this point I’m wildly speculating) is probably trivial to send or embed in some other communication with the server e.g. by ensuring the PRNG is deterministic after the first keygen and faulty in some known way and sending over a future result.

          I wouldn’t trust the French government.