Olvid Messenger

[IkelAtomig]

https://olvid.io/en/ - Seems like this messenger could be in the Real Time Communications page.

[dngray]

Specification: https://olvid.io/assets/documents/2020-12-15_Olvid-specifications.pdf
Source: https://github.com/olvid-io (iOS/Android only)

[ph00lt0]

• Starting a call requires subscription, that wouldn't be as much of a problem, but payments require Google Play Services and payment through Google's in app purchases on android. See no advantage of Signal.
• There is no way to discover users other then via links, or QR codes.
• Crypto seems okay on first look, but needs further investigation.
• App looks and feels very much like a beta version (android)
• The website states that using the app leaves no digital footprint. That is is simply impossible and false.

[realguyman]

Olvid is simply not better than existing solutions (Threema, and Signal). I don't think this should be added to PrivacyGuides.

[ph00lt0]

I do not agree with you here, technically it is actually better then Threema though, as Threema still does not offer Perfect Forward Secrecy, Olvid seems to do so. But yeah Threema is also not listed for it's doubtful security practices.

[finiasz]

I can confirm that Olvid offers Perfect Forward Secrecy. All messages are encrypted using single use encryption keys derived through a "double ratchet" mechanism.

[finiasz]

And Olvid phone calls also offer Perfect Forward Secrecy with end-to-end encryption.

[neorelish]

I tried it to see by myself.
So far so good. yes the UI seems a bit old but everything is working fine.
Compared to others there are indeed some unique features :

• full ephemeral messages and being able to edit/delete them as you wish
• no central user directory (love it): I can choose and authenticate myself each one of my recipient. No more spam (I've received some from fake Amazon in signal)
• no size limit for sending, good for movies ;)
• no phone number or email required, just a cryptographic ID
• Strong security based on independent audit and white paper

It's not perfect and lacks some fun stuff but maybe in the future ? (asking since you seem to be in charge).

Compared to others, I can see why the messenger is suggested. It's not perfect no messenger is.
I think it's a (very) good one, and deserve their seat in the list, IMHO.

PS : yes it would be nice to have free voice call.

[finiasz]

Yes @neorelish, we plan to add some "fun stuff" to the app. It will take some time as our development team is still pretty small compared to some other apps and we already have a lot of features planned in our roadmap!

[finiasz]

Compared to Threema and Signal, the main difference of Olvid is the absence of central user directory. The main drawback (as pointed out by @ph00lt0) is the fact that, in Olvid, you need to "build" your contact list by inviting people to join you on Olvid, but there are many advantages:

• Olvid does not need any personal information to work: users do not need to share their phone number or email address
• Olvid does not require access to your phone contacts
• You will never receive spam on Olvid: people cannot send you messages if you did not deliberately add them to your contacts. I'm pretty sure many of you have received contact requests on Signal advertising for a free iPhone: this cannot happen on Olvid

And regarding the fact that the "App looks and feels very much like a beta version (android)", this is simply a "look and feel" issue and the technology inside Olvid is definitely not in beta!

[ph00lt0]

You will never receive spam on Olvid: people cannot send you messages if you did not deliberately add them to your contacts. I'm pretty sure many of you have received contact requests on Signal advertising for a free iPhone: this cannot happen on Olvid

@finiasz I actually never had this on Signal, and am not familiar with that problem.
May be good to point out that you work for Olvid. (although your responses are welcomed :) )

I think this may have been closed to quickly. I have no experience with this messenger other then my quick try out, but you do seem to take things seriously. There are definitely some things that could be improved on their side such as look and feel. You also seem to have undergone security audits. What mainly holds me from considering this is the requirement for Google Play Services to get a subscription. I am fully aware that for some functionality Signal also has dependence on this (which is criticized by many users) but at least they do not demand it fully for being able to initiate calls.

[finiasz]

Yes, sorry for not clearly stating that I work for Olvid. I am indeed in charge of the Android version of Olvid. But when comparing Olvid with other messaging apps, we always try to be as unbiased as possible and base our arguments solely on facts.

Regarding look and feel, and the UI in general, I 100% agree with you, and some of our users report this to us: when first trying Olvid, the app is not very "visually appealing". But after a while we get used to it! At some point in time we will hire a designer and put some efforts into our UI to be on par with the competition, but for now our efforts have been first on security, and second on features.

Regarding phone calls, we would love to make them part of the free version of Olvid, but for the moment we still rely on in app purchases for this. And rather than implement a payment solution without Google Play Services, we would rather make them free :)

[ph00lt0]

Another interesting fact: one of the co-founders seems to work (or has) at LexisNexis. (see https://olvid.io/team/en/)

[finiasz]

Yes, he used to work there, but this was a long time ago...

[neorelish]

As stated before, I strongly disagree with the selected answer (https://github.com/privacyguides/privacyguides.org/discussions/872#discussioncomment-2502156)

[realguyman]

Just to be clear: I didn't close this discussion, and the “answer” has been redacted. However, it is likely I will not recommend Olvid messenger.

[neorelish]

Since there are a lot of features that other messenger doesn't have, what does it lacks to be recommended ?

[dngray]

@neorelish not sure who you are, but account created 12 hours ago with no public repositories or activity 🤔

[neorelish]

@neorelish not sure who you are, but account created 12 hours ago with no public repositories or activity 🤔

Funny to ask for that on a privacy related site ! :)
I just used to participate in other discussions on Reddit and Matrix but never on GitHub that's all. Anything need a start right ?
Can we be back to the topic ? Just for the record, what is missing and could you point me out to a better alternative with no phone number, ephemeral message and strong authentication ?

[realguyman]

Pros and cons

Pros:

1. No PII (Personally Identifiable Information) is required to operate under the Free plan.
2. No user discoverability (due to privacy preserving nature).
3. Both applications for Android and iOS are open source.
4. Uses a protocol similar to that of Signal's.
5. Olvid's iOS app was audited in early 2020 (PDF). (unfortunately I am not able to review it, I don't understand French.)
6. SAS-based Trust Establishment
   Protocol received positive cryptanalysis in 2020 (PDF).
7. Certification of Security from ANSSI in 2020 for the iOS app (PDF), and 2021 for the Android app (PDF).

Cons:

1. There are no desktop clients, but they're on the roadmap.
2. No video calling, but also on the roadmap.
3. The web client isn't open source.
4. The repository for Olvid's Android application (v0.11.1+build.156 well over a month ago) isn't up-to-date with the latest available version on Google Play Store (v0.12.1+build.161 about two weeks ago).
5. Users are required to use Google Play Store to download the Android application.
6. While not required for the evaluation of Olvid, the backend isn't open source.
7. No user discoverability (due to ease of use for getting in touch with contacts).

---

Why it probably won't be added yet

Signal is a free, reliable, non-anonymous service that is accessible to users on all major platforms (excluding the web, and for good reason). It supports texting, attachments, voice messages, audio calling, and video calling. Signal doesn't provide users with anonymity due to some (ahem, most) countries tying your real identity to your phone number. This can be complicated to work around depending on where you live.

A word on user discoverability. Signal isn't perfect, but it does retain privacy well enough for most users, providing balanced ease-of-use and privacy.

Olvid shouldn't be recommended yet because: it doesn't support video calling, source-available (with legal rights to decompile the program) desktop clients aren't available, payments/downloads are only possible through Google services, and the UX hasn't been great.

It should be noted that depending on users' cases, purchasing a subscription for Olvid could be just as revealing as using your phone number. As for a user ID, obviously it's more privacy preserving to hand strangers a unique ID that isn't PII (which is Signal's severe con).

---

Marketing fluff

DISCLAIMER: Olvid's marketing doesn't affect whether it gets added. Only technical details are considered.

Our security model is utterly game-changing. Olvid is the first and only messaging system whose security no longer relies on any trusted third party, either operators or their servers.

This is far-fetched, because Signal is secure under the given threat model that the servers are compromised, and that has been the case for longer than Olvid. For more information, see the audit conducted on Signal's protocol in 2016 (PDF).

[...] Nor can any users identities ever be revealed. Olvid is the only system that also encrypts metadata, thus guaranteeing the anonymity of interlocutors. [...]

This is false. It's certainly not the only system that encrypts metadata. Signal encrypts metadata and simply doesn't expose it where possible, or within reason.

Inability of the operator to know "who is talking to whom". No third party could ever identify the participants, not even the server. No trace of any metadata.

This is also false. Gathered data (I.P. addresses, receiver identifiers, timestamps, and other data that might be available) can eventually build a profile of “who is talking to whom” on users of Signal and Olvid. But the current countermeasures certainly complicate being able to easily collect a mass amount of information on users.

While this is specific to Signal, these principles can also be applied to Olvid where similar designs are present:

• Improving Signal's Sealed Sender (Video)
• Improving Signal's Sealed Sender (PDF)

As such, it is not easy to claim any messenger to be truly anonymous. Unfortunately, Olvid is no exception, but a step in the right direction.

---

Questions for @finiasz

1. Why isn't the IV authenticated with ciphertext? See: https://github.com/olvid-io/olvid-android/blob/b3838bfc9d0605eb847fec1d87d8fe9041889e90/obv_engine/engine/src/main/java/io/olvid/engine/crypto/AuthEnc.java#L94
2. Why didn't Olvid adopt Signal's protocol?

---

Conclusion

While I believe Olvid has potential, and has brought an interesting design for securely establishing trust, I don't think it's ready to be listed on PrivacyGuides.

[finiasz]

Thanks @realguyman for this detailed explanation. regarding your questions:

1. it's not very clear when we write enc.encrypt(iv, plaintext, ciphertext);, but in fact the IV is part of the ciphertext (ciphertext is longer than plaintext), so the IV is indeed authenticated.
2. We didn't adopt Signal's protocol for two main reasons: there are some elements of the protocol we do not like (for example, the identity of the message sender is required to decrypt, thus the need for a sealed sender: in Olvid's scheme we do not need to know who sent the message to decrypt it, decrypting it reveals who sent it) and we wanted more flexibility to evolve our protocol when needed (typically, when shifting to post-quantum cryptography)

And while I agree with most of your analysis, one element I need to emphasize is that Signal is secure in a model where the central user directory is trusted. The article you point to indeed proves that the channel creation and message encryption is secure, but it assumes that users have the correct public key for their contacts to start with. As far as I know, the only "proof" we have that the directory can indeed be trusted is Intel SGX technology: this is not enough to convince me there is no backdoor in the directory...

[finiasz]

Ah, and we pushed the latest version to Github, sorry for the delay.