3 ms·
To be honest, I don't understand what substantial benefit end-to-end encryption actually brings in an environment of (almost-)mandatory updates. - If someone f
by xg15 11y ago
To be honest, I don't understand what substantial benefit end-to-end encryption actually brings in an environment of (almost-)mandatory updates.
- If someone from Facebook/Telegram/Signal/etc wants to know what you're writing, they can just instruct their app (via update) to send them your key. For closed-source services, you'd theoretically have to decompile and audit each update to make sure they are not doing that.
- If they want to know what you have written in the past, they can instruct the app to send them the conversation log.
- If Google (or Apple or Microsoft, respectively) want to know what you're writing, they can instruct the OS to send them the data.
(Google's "Android Backup Service" for example also backs up "third party settings and data" [1]. I don't know about the details of the backup service, but this shows to me it's quite possible that your key or conversation logs might even land accidentally on some providers' servers without them having any bad intent.)
- If (three letter agency of your choice) wants to get the data, they can just force any of the above companies via NSLs to get it for them.
- If any of the US strategic partners want to get the data, they can likely make a deal with an intelligence agency.
- Lastly, if the messenger company wants to mine or sell user data, they still have a lot of stuff that cannot be encrypted for operational reasons (such as your contact list and the phone numbers of all your contacts).
That leaves to me the only group for which "overlay encryption" brings an actual benefit political activists in a country not at all affiliated with the US - or highly knowledgeable individuals who carefully control which updates they get. Both groups are important to consider but likely had ways to protect their communication before.
To actually protect communication not just from "the government" but also from the private industry, we would at least need some independent party to vet app updates.
[1] https://support.google.com/nexus/answer/2819582?hl=en https://support.google.com/nexus/answer/2819582?hl=en
- rashkov 11y agoI think the assumption is that any sufficiently motivated attacker will find a way to compromise their target. What e2e encryption accomplishes is the end of dragnet surveillance of entire societies, something that Snowden has exposed in great detail already.
- xg15 11y agoBut my point is that the way end-to-end encryption is implemented currently doesn't even accomplish that: Either you trust your messenger provider not to be complicit in any surveillance or data mining activities, then plain TLS (with key-pinning) is enough to save you from being snooped - or you don't trust them, then you're not actually protected as they could push an update at any time to sidestep the encryption. There are certain cases were current end-to-end encryption brings more security - e.g. if a provider's data center is compromised but the provider itself isn't. But those seem like edge cases to me that don't justify the attention the feature is currently getting.
- pfg 11y agoIt's not perfect, but E2E crypto does make it significantly harder to get to the plaintext. A backdoor will always leave some kind of paper trail (if you know what to look for), whereas intercepting the plaintext on the server cannot be detected by users at all. Legally speaking, compelling a company to intercept messages that they have access to sounds different than compelling them to develop and sign a backdoored version of their product - that's essentially the Apple vs. FBI case. I think this is a case of perfect being the enemy of good. If a state actor wants access to your messages, you're probably screwed anyway, unless your OpSec is top-notch. E2E crypto means that they'll have to really want your messages, at which point they're probably better off just stealing your phone or using a $5 wrench.
- antihero 11y agoThe key is in the penultimate question. If GCHQ/NSA is interested in surveilling a specific individual, you'd need a lot more than this software to communicate securely, because there are just so many vectors. This solves dragnet surveillance, where the government could potentially just sniff all communication and use various analytical techniques to sniff out potential criminals or people they dislike.