4 ms·

It's annoys me that both sides lie/are wrong in this. The statement' signatories [1] agree that having default E2EE (end to end encryption) on Facebook Messeng

It's annoys me that both sides lie/are wrong in this. The statement' signatories [1] agree that having default E2EE (end to end encryption) on Facebook Messenger will stop around 12 million reports of child pornography per year. This is a lie/incorrect because recipients of such material can manually report it [2] and/or it's possible to make the detection tool work on end points, it's a fancy hash comparison algorithm [3]. The Matrix people respond to what they think "lawful access" is by saying that "lawful access " is a EE2E backdoor so they it's impossible to have "lawful access" and "E2EE". It's an assumption that "lawful access" is a third decryption key and it's possible to have E2EE and third party access to it as Facebook Messenger proves [2]. If all parties were honest the governments would say "we want regular police work to have a way to lawful access E2EE messages and while you're a it a backdoor for the intelligence agencies". The other side would have positions between "no to both" but most would say "no backdoor, OK to past encrypted messages for police investigations and intelligence agencies with transparency reports and oversight". [1] https://www.gov.uk/government/publications/international-statement-end-to-end-encryption-and-public-safety/international-statement-end-to-end-encryption-and-public-safety-accessible-version#signatories https://www.gov.uk/government/publications/international-sta... [2] https://www.facebook.com/help/messenger-app/android/498828660322839?rdrhc https://www.facebook.com/help/messenger-app/android/49882866... [3] https://en.wikipedia.org/wiki/PhotoDNA https://en.wikipedia.org/wiki/PhotoDNA

robertlagrant 6y ago
Having a button which people can use to forward a message (or forward it) - with a cryptographic guarantee of who it came from - to the authorities would be useful. E2EE preserved, and in fact utilised.
- cyphar 6y ago
  Signal and other OTR-derived systems actually cannot cryptographically prove that someone sent a message (the recipient can forge the message because the message is authenticated via HMAC not a signature).
  - Arathorn 6y ago
    Yup. Matrix/Olm/Megolm similarly has deniability. You can prove in retrospect that someone in the conversation sent the message, but not who.
    - cyphar 6y ago
      To be honest, I'm not convinced that this is practically true in Matrix except in cases where you wish to argue that the homeserver operator is "in on it" to create fraudulent messages (the same goes for Signal -- though in Matrix's case the homeserver usually keeps the entire conversation history, making it even harder to make that argument). So while it is technically true that you cannot cryptographically prove that a message was sent by an individual, practically speaking you probably can prove it.
wffurr 6y ago
What's the distinction between "lawful access E2EE messages" and "backdoor for the intelligence agencies"? How is the service provider supposed to ever access an E2EE message without a back door?
- warkdarrior 6y ago
  You are confusing policy with mechanism. There is no good solution right now for lawful access to E2EE messages, but the blanket opposition to it prevents are work on finding such a solution.
  - wffurr 6y ago
    The two are inextricably intertwined. Only the sender or recipient can provide lawful access to an end-to-end encrypted message. That's the whole point of end-to-end encryption. e.g. an informant can take screenshots or copy the message text or a defendant can be compelled to provide their fingerprint by subpoena.
  - cyphar 6y ago
    There is no solution because by definition any solution must be some kind of backdoor -- the explicitly stated goal is to allow retrospective access to encrypted communication without the consent or knowledge of the parties involved. To me, that is almost a textbook definition of a backdoor in an encrypted communication protocol. The blanket opposition isn't out of stubbornness and lack of understanding, it's because there is no way to simultaneously satisfy people who want a backdoor in every E2EE communication protocols and those who don't.
- tumetab1 6y ago
  The service provider, in most cases, is also the application provider which decrypts the message and holds the keys. Most "lawful access" scenarios are met if the service/application provider requests/obtains the plain text and/or keys from the application. Example: Police is investigation John Doe for involvement in child porn. Police has his phone number so requests a warrant for his chat apps message contents. Gets a warrant so sends a request to Facebook. Facebook sends request to all apps to send last month decrypted content. Facebook provides that to the police.
exo762 6y ago
That's a distinction without a difference. Why would anyone run closed-source clients for federated networks when there are open-source clients readily available? One could imagine a scheme where endpoints run some kind of zksnark-wrapped computation against a hash-list of "unlawful" content, proving with mathematical certainty that one is not transmitting contents that are on the list. But this mountains of effort both because of complexity (protocol needs to be structured in some other way) and limitations of snarks - they do not handle this amount of computation well.
- tumetab1 6y ago
  I'm not fully aware how Matrix works, my comment is more related to their statement trying to make equal "lawful access" and an E2EE backdoor. Why would anyone run closed-source clients for federated networks when there are open-source clients readily available? It doesn't have to be closed-source, it can be open-sourced and still have the functionality and send the plain text to the server upon request. Why would anyone spend time implement it? Well, that could be defined in the law that for open-source the government would have to provide the pull request for it a fork. Why would criminals use such software? Because many criminals aren't that operational-security aware so they use SMS, Facebook Messenger and other easily intercepted messaging software/channels. A popular open-sourced software with E2EE decryption mechanism would still be used by criminals just because is popular.

robertlagrant 6y ago

Having a button which people can use to forward a message (or forward it) - with a cryptographic guarantee of who it came from - to the authorities would be useful. E2EE preserved, and in fact utilised.

cyphar 6y ago
Signal and other OTR-derived systems actually cannot cryptographically prove that someone sent a message (the recipient can forge the message because the message is authenticated via HMAC not a signature).
- Arathorn 6y ago
  Yup. Matrix/Olm/Megolm similarly has deniability. You can prove in retrospect that someone in the conversation sent the message, but not who.
  - cyphar 6y ago
    To be honest, I'm not convinced that this is practically true in Matrix except in cases where you wish to argue that the homeserver operator is "in on it" to create fraudulent messages (the same goes for Signal -- though in Matrix's case the homeserver usually keeps the entire conversation history, making it even harder to make that argument). So while it is technically true that you cannot cryptographically prove that a message was sent by an individual, practically speaking you probably can prove it.

wffurr 6y ago

What's the distinction between "lawful access E2EE messages" and "backdoor for the intelligence agencies"? How is the service provider supposed to ever access an E2EE message without a back door?

warkdarrior 6y ago
You are confusing policy with mechanism. There is no good solution right now for lawful access to E2EE messages, but the blanket opposition to it prevents are work on finding such a solution.
- wffurr 6y ago
  The two are inextricably intertwined. Only the sender or recipient can provide lawful access to an end-to-end encrypted message. That's the whole point of end-to-end encryption. e.g. an informant can take screenshots or copy the message text or a defendant can be compelled to provide their fingerprint by subpoena.
- cyphar 6y ago
  There is no solution because by definition any solution must be some kind of backdoor -- the explicitly stated goal is to allow retrospective access to encrypted communication without the consent or knowledge of the parties involved. To me, that is almost a textbook definition of a backdoor in an encrypted communication protocol. The blanket opposition isn't out of stubbornness and lack of understanding, it's because there is no way to simultaneously satisfy people who want a backdoor in every E2EE communication protocols and those who don't.
tumetab1 6y ago
The service provider, in most cases, is also the application provider which decrypts the message and holds the keys. Most "lawful access" scenarios are met if the service/application provider requests/obtains the plain text and/or keys from the application. Example: Police is investigation John Doe for involvement in child porn. Police has his phone number so requests a warrant for his chat apps message contents. Gets a warrant so sends a request to Facebook. Facebook sends request to all apps to send last month decrypted content. Facebook provides that to the police.

exo762 6y ago

That's a distinction without a difference. Why would anyone run closed-source clients for federated networks when there are open-source clients readily available? One could imagine a scheme where endpoints run some kind of zksnark-wrapped computation against a hash-list of "unlawful" content, proving with mathematical certainty that one is not transmitting contents that are on the list. But this mountains of effort both because of complexity (protocol needs to be structured in some other way) and limitations of snarks - they do not handle this amount of computation well.

tumetab1 6y ago
I'm not fully aware how Matrix works, my comment is more related to their statement trying to make equal "lawful access" and an E2EE backdoor. Why would anyone run closed-source clients for federated networks when there are open-source clients readily available? It doesn't have to be closed-source, it can be open-sourced and still have the functionality and send the plain text to the server upon request. Why would anyone spend time implement it? Well, that could be defined in the law that for open-source the government would have to provide the pull request for it a fork. Why would criminals use such software? Because many criminals aren't that operational-security aware so they use SMS, Facebook Messenger and other easily intercepted messaging software/channels. A popular open-sourced software with E2EE decryption mechanism would still be used by criminals just because is popular.