4 ms·

Govt don't need to break encryption technically. They just need to have lawful orders issued to the app or platform companies to comply to provide the informati

Govt don't need to break encryption technically. They just need to have lawful orders issued to the app or platform companies to comply to provide the information about the targets (1-1 chats or group chats). Question is - can the companies comply? Do the companies themselves have backdoors so they can comply? Btw, there are laws on the books that say they ought to have mechanisms such that they can comply.

MattPalmer1086 3y ago
The only way a company providing end to end encrypted messaging service can comply with such an order is to ship a backdoored version of their client software. This might send copies of the messages somewhere else as well, or use deliberately weakened keys, or something like that. There is no way for the companies to break the encryption on their servers without compromising the client first.
- vinay_ys 3y ago
  Usually, each message is encrypted with a new symmetric key that's unique per message and that key is encrypted with the public key of the recipient and included in the envelope of the message. It isn't hard to ask the client to also encrypt the symmetric key to company's own monitoring public keys and include it in the envelope. Presumably this can be done dynamically on-demand (when a govt monitoring request comes in) by the company servers. It is also conceivable that all symmetric keys of past 30 days of messages are stored on the client and can be asked to be sent to server on-demand. If all this functionality ("backdoor") is already baked into the client, and carefully masked to look like legitimate functionality of the app, then there's no need to "compromise" the client afterwards. We are talking about private, for-profit companies running closed-source service with closed-source client and with TLS-key-pinned client-server RPCs. And we are talking about closed-source phones running closed-source hardware, firmware, OS, and platform services. Even if the app was open-source (like Signal) there are plenty of injection points to do the above at the lower-levels.
  - MattPalmer1086 3y ago
    Yes, there are many ways it could be done. However, I doubt that a private company whose business model was end to end encryption would secretly bake a back door into every client by default. Any exposure of this by reverse engineering or whistle blowing would end the business. Why would a private company work against their own interests in that way? I could imagine that specific clients got special back doored versions under some kind of secret court order, but I'm not sure what the delivery mechanism would be for that. Edit: your final point that you could compromise a device independently of the app itself seems much more likely to me. Tools to do this exist and have been provided to governments.

MattPalmer1086 3y ago

The only way a company providing end to end encrypted messaging service can comply with such an order is to ship a backdoored version of their client software. This might send copies of the messages somewhere else as well, or use deliberately weakened keys, or something like that. There is no way for the companies to break the encryption on their servers without compromising the client first.

vinay_ys 3y ago
Usually, each message is encrypted with a new symmetric key that's unique per message and that key is encrypted with the public key of the recipient and included in the envelope of the message. It isn't hard to ask the client to also encrypt the symmetric key to company's own monitoring public keys and include it in the envelope. Presumably this can be done dynamically on-demand (when a govt monitoring request comes in) by the company servers. It is also conceivable that all symmetric keys of past 30 days of messages are stored on the client and can be asked to be sent to server on-demand. If all this functionality ("backdoor") is already baked into the client, and carefully masked to look like legitimate functionality of the app, then there's no need to "compromise" the client afterwards. We are talking about private, for-profit companies running closed-source service with closed-source client and with TLS-key-pinned client-server RPCs. And we are talking about closed-source phones running closed-source hardware, firmware, OS, and platform services. Even if the app was open-source (like Signal) there are plenty of injection points to do the above at the lower-levels.
- MattPalmer1086 3y ago
  Yes, there are many ways it could be done. However, I doubt that a private company whose business model was end to end encryption would secretly bake a back door into every client by default. Any exposure of this by reverse engineering or whistle blowing would end the business. Why would a private company work against their own interests in that way? I could imagine that specific clients got special back doored versions under some kind of secret court order, but I'm not sure what the delivery mechanism would be for that. Edit: your final point that you could compromise a device independently of the app itself seems much more likely to me. Tools to do this exist and have been provided to governments.