So...TrueCrypt is dead. Canary warrant?

Check out:

Er, warrant canary not the other way around.



Conspiracy theorists are speculating it has to do with the Snowden interview about to air.

Perhaps Schneier will have something to say about it.

5 hours since his last post, tweet.

Is it possible that it simply means that the shadowy folks behind Truecrypt want people to stop using it, since it hasn’t been in active development for a long time?

Sounds like someone got a NSL.

This is so bizarre that is has to be either a hack or a warrant canary. My money is on warrant canary or similar reaction to pressure by the devs, because if that was a hack it’s a lot of effort for a fairly poor payoff. Not particularly funny or liable to fool anyone.

Nobody really knows what happened. The truecrypt authors are anonymous. If the government tracked down their identities and sent a NSL, they could have just said “we are no longer developing this, someone fork it”. Instead they went to the trouble to change the code to decrypt-only, and wrote up that long page telling people to use bitlocker. The truth is that nobody really knows what the deal is.

Absolutely true, all we can do is speculate. And avoid “upgrading” to TrueCrypt 7.2.

At my place of employment we are required to use TrueCrypt full disk encryption for all our laptops. I’m painfully tempted to send an email to our CTO tomorrow that I upgraded my TrueCrypt and it doesn’t seem to be working right. Just to start a fire drill.

Is it just me that trusting encryption to a faceless group that may have been compromised by the government seem somehow ironic?

Brian Krebs says it’s apparently legit. Most likely reason: the recent security audit revealed code quality so terrible that fixing it would have amounted to a total rewrite, and the unpaid (one-person?) team couldn’t be bothered. See this comment:

The iSec initial audit report was very critical of the TC code quality, and implied that it looks like the work of a single coder. There was no update for 2 years. The build process requires a 20 year old MS compiler, manually extracted from an exe installer.

After OpenSSL that’s the second big open-source security meltdown where widely trusted software turned out to be terribly crummy, once someone actually bothered to look at it. Trust no one!

OpenSSL was actually insecure; as far as we know truecrypt isn’t. OpenSSL had poor code quality because the developers were paid to meet government requirements. We have no clue who developed truecrypt. Very different situations.

The lesson from all this is that open source doesn’t necessarily mean secure, even for long-standing software.

I don’t think it’s as easy as “they could just say” – there’s that XKCD comic that shows a hypothetical situation where evildoers (government or otherwise) don’t get stymied by fancy encryption but instead use a $5 wrench to coerce the key.

Other people have noticed weird little things in source like changing of US to United States. The whole “out there” solution of recommending a switch to BitLocker (with MS already likely in bed with the NSA) is just one potential tipoff that it’s a warrant canary.

Krebs is just making his best guess too at this time.

Can you clarify/elaborate for those that are curious?

Random, mostly anonymous online strangers aren’t an infallible panacea? Inconceivable! Next you guys will be telling me Wikipedia isn’t 100% accurate.

I’d hazard a guess that usually government code requirements emphasize old standards and ideas that permit or even require coding that’s just up to today’s standards (ie. lags behind). Or, to the cynical, means it could easier for them to break and/or bloat.

— Alan

The US government has a certification program called “FIPS”, which is extraordinarily strenuous to meet. It controls which encryption standards are supported, how they’re implemented, validation, etc. Many of those standards are useless/obsolete/broken and/or positively ancient. Beyond encryption standards that could be brute-forced with your cellphone, OpenSSL still supported DOS, Win16, and pre-OSX MacOS. Some, but not all, of those same constraints are shared by other governments.

The OpenSSL foundation, which paid the developers’ salaries, was largely funded by commercial organizations that needed to comply with FIPS. Their time was spent largely in ensuring FIPS compliance, because that’s what paid their bills. But it just barely paid those bills, OpenSSL was poorly funded, so the reality is they spent most of their time working on FIPS compliance rather than actually improving OpenSSL.

The goodbye message explicitly states the possibility of unfixed security issues as a reason to migrate away from it. Unless you know for sure that’s all made up I should take the developers’ word for it.

Regarding the warrant canary idea, why would the US government (or any other) serve TrueCrypt with a warrant? The organization does not store other people’s data, as with Lavabit and other service providers that have been targeted by the NSA, and the source code was already available for download. There’s literally nothing the developers have that the government doesn’t.

Given the recent audit with its harsh criticism of code quality and ancient development tools, I think the most straightforward explanation is the most likely: the developer(s) anticipated having to rewrite the whole thing and couldn’t be bothered. It’s already been dormant for years, after all.

Consensus – really speculation fwiw considering no one admits to knowing a dev – seems to be the code was just so bad that it couldn’t be readily fixed. Pity if true. Sensible coding standards, methods and procedures aren’t really all that hard to implement, open source or not, so long as the developers buy into the idea. And what kind of project requires higher standards than encryption? for Windows users now, I guess.