Google's E2Email Gmail Encryption Looks a Lot Like Vaporware(wired.com)
wired.com
Google's E2Email Gmail Encryption Looks a Lot Like Vaporware
https://www.wired.com/2017/02/3-years-gmails-end-end-encryption-still-vapor/
6 comments
I am not surprised. I got sick in 2014 and spent a summer implementing OpenPGP from scratch. It is _extremely_ ugly. If I was Google I would make something better and push that, instead of spending tearful days implementing the really bad specification that is RFC4880.
I feel like there is a logical connection between you getting sick and working on an OpenPGP implementation, but I cannot for the life of me figure out what that connection is.
Take a look at https://cryptup.org (disclaimer - my project). Encryption can be made easy.
RFC4880 may be ugly. Yet great design can still turn it into a good user experience.
I don't think anyone considers TCP/IP pretty. Who cares? When I open my browser, I see none of it.
RFC4880 may be ugly. Yet great design can still turn it into a good user experience.
I don't think anyone considers TCP/IP pretty. Who cares? When I open my browser, I see none of it.
Well, given that Google is giant, unbelievably powerful, and have a very significant share of email market - they could've pulled just about anything. If OpenPGP isn't good (and it's true that it has issues), they could've implemented and proposed a whole new standard that addresses existing shortcomings, evangelize/market it - and it's very likely others would've followed.
Mailvelope is OpenPGP as a browser extension and works ok with GMail:
https://github.com/mailvelope/mailvelope
https://github.com/mailvelope/mailvelope
So, it is still decidedly beta (more accurately, alpha), but I wrote and just released http://gibber.it, to allow users to send end-to-end encrypted messages through basically any place in a browser that you can enter text.
I wrote this because it seems to me that the lack of end to end encryption in services like gmail is a problem - it also seems to me that there is no reason to have this service tied to a particular website, or for it to be administrated by another service provider. Accordingly, all GibberIt does is provide the encryption and key exchange - use it on gmail, nytimes comment boards and reddit (all working well so far) - or wherever the hell else you want. It will soon (I hope...) be working on facebook too (their content-security-policy is very strict - rightly so - and I am making the extension compatible with these requirements).
* It currently functions as a chrome extension.
* Sign up, invite connections just like any other social network.
* Encryption is end to end, AES 128 with nonce'd salts.
* Use a password you share with your connection (NOT your login password) to send connection invites - this is used to encrypt your keys during the invite process. (Make sure to accept the return invite! This is how your connection sends his or her keys back to you. Also note that you will likely need to reload any tab running the extension after accepting an invite in order to get the keys to load.)
* Use the chrome extension to encrypt and decrypt messages as you browse.
* I am, by profession, a corporate, software and information lawyer. More about me here: http://www.lawyernamedliberty.com
Demonstration gifs of GibberIt here: https://www.gibberit.com/#!how
Please note that the system is in BETA. Still many tweaks to work out. Use is at your own risk.
Please feel free to ask any questions you may have. I welcome any and all feedback. Love the system? Hate the system? Please let me know.
Edit: Please note that the gibberit homepage - AND NO OTHER PAGE - uses google analytics. This is clearly detailed in my privacy policy. Aside from that, I do not use any tracking software.
I wrote this because it seems to me that the lack of end to end encryption in services like gmail is a problem - it also seems to me that there is no reason to have this service tied to a particular website, or for it to be administrated by another service provider. Accordingly, all GibberIt does is provide the encryption and key exchange - use it on gmail, nytimes comment boards and reddit (all working well so far) - or wherever the hell else you want. It will soon (I hope...) be working on facebook too (their content-security-policy is very strict - rightly so - and I am making the extension compatible with these requirements).
* It currently functions as a chrome extension.
* Sign up, invite connections just like any other social network.
* Encryption is end to end, AES 128 with nonce'd salts.
* Use a password you share with your connection (NOT your login password) to send connection invites - this is used to encrypt your keys during the invite process. (Make sure to accept the return invite! This is how your connection sends his or her keys back to you. Also note that you will likely need to reload any tab running the extension after accepting an invite in order to get the keys to load.)
* Use the chrome extension to encrypt and decrypt messages as you browse.
* I am, by profession, a corporate, software and information lawyer. More about me here: http://www.lawyernamedliberty.com
Demonstration gifs of GibberIt here: https://www.gibberit.com/#!how
Please note that the system is in BETA. Still many tweaks to work out. Use is at your own risk.
Please feel free to ask any questions you may have. I welcome any and all feedback. Love the system? Hate the system? Please let me know.
Edit: Please note that the gibberit homepage - AND NO OTHER PAGE - uses google analytics. This is clearly detailed in my privacy policy. Aside from that, I do not use any tracking software.
How would they display ads in email accounts if the content is encrypted? Just pure pragmatic reason.
Search would be difficult as well though.
Even if that were true, that's not reason enough for me. Why can't they just display random ads on the side-bar? There's no reason they need to read my emails for that.
Because anonymized ads cost advertiser wayyyy less than targeted ones. so gmail profits would fall significantly
profits falling != impossible to do
There's no reason they couldn't still send targeted ads based on other user info.
This is a choice gmail has to make between doing the right thing, and doing the profitable thing.
There's no reason they couldn't still send targeted ads based on other user info.
This is a choice gmail has to make between doing the right thing, and doing the profitable thing.
Google stopped working on End-to-End at least a year ago:
https://motherboard.vice.com/en_us/article/google-yahoo-end-...
And now they've "open sourced it" with minimal changes since then. Sounds to me like they stopped working on it, but they didn't want to keep getting asked by the tech media and crypto people about the project's status. So they went "Here, it's all yours now! You're welcome. Yes, we know how awesome we are."
But it doesn't look quite finished, so I wonder if anyone will bother to complete Google's half-assed job. Google didn't even bother to integrate it with Key Transparency before open sourcing it. That speaks volumes as well.
They've recently also released a centralized/hosted version of S/MIME for enterprise users as well, again showing their complete disinterest in continuing to support end-to-end crypto.
https://security.googleblog.com/2017/02/hosted-smime-by-goog...
I'm guessing they stopped working on End-to-End around the time they started working on the hosted S/MIME. I wouldn't get my hopes up about Google supporting end-to-end crypto in the future. Now I'm starting to wonder if they'll also kill the feature in Allo, or just hide it some more in future releases, so that it becomes even harder to use.
https://motherboard.vice.com/en_us/article/google-yahoo-end-...
And now they've "open sourced it" with minimal changes since then. Sounds to me like they stopped working on it, but they didn't want to keep getting asked by the tech media and crypto people about the project's status. So they went "Here, it's all yours now! You're welcome. Yes, we know how awesome we are."
But it doesn't look quite finished, so I wonder if anyone will bother to complete Google's half-assed job. Google didn't even bother to integrate it with Key Transparency before open sourcing it. That speaks volumes as well.
They've recently also released a centralized/hosted version of S/MIME for enterprise users as well, again showing their complete disinterest in continuing to support end-to-end crypto.
https://security.googleblog.com/2017/02/hosted-smime-by-goog...
I'm guessing they stopped working on End-to-End around the time they started working on the hosted S/MIME. I wouldn't get my hopes up about Google supporting end-to-end crypto in the future. Now I'm starting to wonder if they'll also kill the feature in Allo, or just hide it some more in future releases, so that it becomes even harder to use.
E2E email is a bit of a lost cause, so probably better that they not waste their time.
why do you say that?
It is extremely hard to use compared to Signal or Whatsapp and has worse cryptographic properties.
I highly disagree. Additionally, I'm not convinced Signal or Whatsapp in their current conditions viable replacements.
Counterexample to "hard to use" point: Protonmail.
(Arguably, it's easier to use for some scenarios, compared to Signal or WhatsApp, because one doesn't need to have a reachable mobile phone number.)
As for the crypto - I think, OpenPGP isn't broken or "worse", just has different properties (like lack of standardized PFS, which has both cons and pros). But, yes, protocol limitations for metadata security (e.g. subject lines) are an issue.
(Arguably, it's easier to use for some scenarios, compared to Signal or WhatsApp, because one doesn't need to have a reachable mobile phone number.)
As for the crypto - I think, OpenPGP isn't broken or "worse", just has different properties (like lack of standardized PFS, which has both cons and pros). But, yes, protocol limitations for metadata security (e.g. subject lines) are an issue.
Problem with protonmail is that it is only end to end encrypted within their system...
It's not the only problem they have (there are a lot of important functionality missing or limited), but it's not related to "extremely hard to use" point, and it's all doable and isn't done because they have limited resources.
That Motherboard article doesn't indicate that Google stopped working on the project, merely that progress is very slow.
Does client side S/MIME (which is E2E) work in Gmail/Google Apps provided you have a client that supports it (e.g. Outlook)?
If so, I'm not sure what the problem is with them not officially supporting it other than the awkwardness of dealing with certificates.
If so, I'm not sure what the problem is with them not officially supporting it other than the awkwardness of dealing with certificates.
Yes it does work.
Chrome apps are themselves deprecated. This looks dead on arrival.
Hosted S/MIME is entirely different. It's completely unrelated to their general E2E efforts. The enterprise S/MIME solution is mostly to sell security to enterprises, not privacy to individuals.
Hosted S/MIME is entirely different. It's completely unrelated to their general E2E efforts. The enterprise S/MIME solution is mostly to sell security to enterprises, not privacy to individuals.
Google once again has proved that 'Google fighting surveillance' sounds like 'rock musicians fighting drugs' and 'bees fighting honey'.
aside from lack of funding, it might actually be better for the community to maintain this, no? I'd certainly prefer an independent set of developers build this, as principle.
PS: goog employee
PS: goog employee