WhatsApp AES decryption key for all chats leaked | Hacker News Reader

WhatsApp AES decryption key for all chats leaked

joindiaspora.com

110 pointsTeMPOraL39 comments

I wonder if this is similar to Snapchat. Snapchat uses AEC-ECB encryption (which you should never use) with a single key that is not-so-carefully hidden in every binary file they've ever distributed to a mobile device. The kind folks at Gibsonsec revealed Snapchat's "security by obscurity" and documented the API so well that I was able to make a Java Snapchat client in a few hours.

I hope that's not the case for WhatsApp, they have a lot more to lose.

Is this just a key to encrypt chat logs local to the phone? I'm assuming they're not using this to encrypt messages in transit...

If it's just local storage, then obviously it'd be easy for anyone to figure out how the app reads the logs. 95% or more of chatting programs and apps do not even encrypt chat logs in the first place.

Not sure. Snapchat uses that in transit.

What does YCombinator have to lose?

What does HackerNews have to lose?

Why has this story been killed so it doesn't appear on the front page?

I feel there is technical merit in discussing the poor security practice of WhatsApp, so that others don't copy their example.

> Why has this story been killed so it doesn't appear on the front page?

Looks like it's back, for now.

What's also interesting about this is to know that Diaspora still exists.

That's honestly all I got out of this.

mission accomplished, probably.

Somewhat ironically, Diaspora itself has had a staggering number of critical vulnerabilities found since its inception. Numerous persistent XSS flaws were found in the first release.

Is this new news? I thought there were already tools out there making use of this key (eg a proof of concept app to steal the whatsapp db on an android [1])

[1] https://news.ycombinator.com/item?id=7380136

The AES key in that post appears to be:

346A23652A46392B4D73257C67317E352E3372482177652C

The one in the linked Tweet appears to be different.

Sigh You can't 'leak' a symmetric key. If the messages are encrypted on your phone then you already have the key. If they're not then it doesn't matter.

That it uses the 'same' symmetric key for all users is what has everyone concerned. Coupled with the fact that WhatsApp stores your chats on external SD cards [1] without access controls, any APP you have installed can access and decrypt your WhatsApp chat history using this key.

[1] http://bas.bosschert.nl/steal-whatsapp-database/

This was on HackerNews a few days ago https://news.ycombinator.com/item?id=7380136

The security flaw here is that there aren't any access controls, not that they use the same symmetric key everywhere for local storage. Even if they store the logs in plaintext, other apps should not be able to read those logs.

"true", but maybe the key was obfuscated... Anyways, it seems that WhatsApp didn't encrypt messages until 2012, then they started using symmetric crypto http://pastebin.com/g9UPuviz

If confirmed, this is a huge blow. It's not just about regenerating the leaked key -- a design relying on a single master key for multiuser messaging is fundamentally broken. It will take months to rebuild it in a sane way.

It has been known from day 1 that Whatsapp "encryption" uses this standard key. This isn't even news.

In terms of huge blows to WhatsApp, nothing beats the original[1] leak which showed that WhatsApp was using IMEIs as a password.

That didn't stop WhatsApp from growing into the behemoth (that its acquisition price states) it is.

[1] http://samgranger.com/whatsapp-is-using-imei-numbers-as-pass...

hmm, this guy has a lot of followers on Twitter, and this tweet is a couple days old. The lack of shitstorm, I presume means this isn't all it's cracked up to be? What is it that is supposedly decryptable with this key? Without more info I'm kinda stumped as to what we're looking at.

He's the main author of cryptocat, which in the past has gotten allot of flak for it's javascript/client side implementation of encryption.

Yeah, I was actually a bit surprised that this thread wasn't mostly people pointing out the irony of this particular person scoffing at someone's security flub.

This is the nature of shipping client code. Any key they used can be extracted by a sufficiently determined individual. The real security issue was not storing the database in the app's private directory.

The real security issue is using the same symmetric key with every client.

That's really just security through obscurity. You're still shipping code that explains the process of obtaining the key. If they store their credentials to obtain the key in a public directory, it's just as vulnerable.

Wrong. If a phone can only retrieve the key for it's own number (e.g. via SMS request), that's orders of magnitude better than the current case where one key can decrypt logs for any arbitrary number. Each SMS request could generate a new key, so even if another app on the same phone does it, it won't be able to get the key to read the logs.

This still relies on the legitimate app not storing the key they fetched in a public directory so the attacker can read it. You can keep adding layers upon this, but it doesn't change that.

>This still relies on the legitimate app not storing the key they fetched in a public directory so the attacker can read it.

Obviously. Why would they do anything else? The point is that they can safely store the logs on an SD card under space constraints.

Not really. No matter what or how many keys they used, someone who stole your device could root your phone and determine the key.

Unless they did something like require the user to enter a master passphrase every time they started the app, then there is no real way around this.

The two day old source tweet: https://twitter.com/kaepora/status/445623864065007616

So it this a newer version of the old key from this study?

http://blog.digital-forensics.it/2012/05/whatsapp-forensics....

Or is this the key they use to transmit the data with?

Link to a screenshot of a tweet citing no source, posted on a dead social network.

Why bother with encryption if you're going to do this?

"All your chats are encrypted with the Advances Encryption Standard."

Which is military grade! :)

Because it's low cost and better than nothing.

I'm not defending their approach, I think it's terrible, I'm just saying that the cost of simetrically encrypting a file with a hard-coded key in Android is too low not to do it...but you should improve it ASAP...

It uses the same key for all instances? Wow.

Could you use some kind of master password that is entered temporarily to encrypt/decrypt the AES decryption key?

Yes. That is, for example, how Firefox stores passwords (if you enable the master password option).

I imagine they haven't bothered to do this for usability reasons, since people would have to type in the password every time they start up the app. Adding it as an option certainly couldn't hurt, but even then I'm not sure there's a point. Most people leave their phones on most of the time, and WhatsApp is probably going to be running in memory that whole time, in which case someone who steals your phone will still probably be able to read all your chat logs anyway.

Wow, i think Diaspora is cool. Is there any HN related thread about Diaspora?

Reply on news.ycombinator.com