Hacker News Reader

Comment by squarefoot | Hacker News Reader

Plausible deniability. The system should allow two (or more) passwords, one unlocks only the important stuff and one unlocks much less dangerous stuff while destroying any evidence of the first, including the multiple passwords protection layer. To add some credibility, the less dangerous data should contain something one could get in trouble for but not enough to have his life destroyed. As an example, if after being forced to reveal a password my phone showed photos of myself having sex with a hooker, I don't think many prosecutors would believe I put them there to protect other more sensitive stuff that would ruin my entire life and not just my relationship. And if they did, I just served them with real proof of why I encrypted my phone which should be enough to counter their unfounded accusations.

View on HN

Plausible deniability systems are pretty amazing but didn't seem to catch on. Truecrypt was awesome, it had hidden drives.

Edit: "didn't seem to catch on" I mean you can't do this with popular products like FileVault.

https://www.truecrypt71a.com/documentation/plausible-deniabi...

If LEO find you're using Truecrypt, do you think they'll not realize that you might have a hidden drive and ask for that password too??

That is where LEO is screwed. They can't prove if you do or you don't have hidden data. EG. There isn't a /hidden_stuff_here/ directory. Its somehow baked into the encryption blocks.

Well, then you give them your second password, and it is fine, because you actually have 3 passwords.

The police can't just keep saying "we didn't find anything, therefore you MUST have ANOTHER double secret password!".

At that point you are indistinguishable from someone who is innocent.

The device is rated 1TB. Given all of the passwords you gave us we can account for .5TB and we cannot find a free-space map. Ergo there is at least one more password that you're not sharing with us.

Seriously, other people aren't dumb.

That's not how partions work for these things. And the people who wrote these security systems thought of exactly this attack vector.

If you create a passworded partion, it adds a bunch of extra unused space, that MIGHT be another passworded partion, or might just be random empty space.

IE, maybe you have 3 passwords and 200 GB of extra space, or maybe it is actually 4 passwords and 150 GB of extra space that is random bytes.

These 2 situations are identical and cannot be differentiated.

The police tell you to access that 150GB of extra space, and it is impossible. Because it's random bytes, and there really isn't an extra partition.

You're missing the point, which is that LEO and courts will (reasonably!) make the following inference:

  "MIGHT be another password-protected partition"
    ->
      "is a password-protected partition"

You can disagree with that thinking all you want, but it won't do you any good if the people thinking that are in control of your destiny.

Sure, but then nothing you do matters.

Imagine the situation where you really ARE innocent. Jail for life? Possible, but whether you are innocent or not doesn't change anything.

Yes, the police could just jail you for life or kill you. But that's ALWAYS the case. They'd do that to you anyway of you are innocent

Well, there is something you can do: don't use software that gives you plausible deniability.

BTW, someone posted this fascinating link in this thread: https://web.archive.org/web/20100820175505/http://iq.org:80/...

Partition will report it’s 1TB large[0]. You should perhaps read about the software before continuing to post comments claiming it’s unfit for purpose. There is a list of considerations that a TrueCrypt/VeraCrypt user with stricter security requirements would have to keep in mind, following all of it makes it fairly difficult to discover hidden partitions. The people who wrote and maintain the software aren’t dumb either.

[0] https://security.stackexchange.com/a/58150

> The police can't just keep saying "we didn't find anything, therefore you MUST have ANOTHER double secret password!"

You might want to watch "Making a Murderer". Or, for the UK equivalent, have a look at the Cleveland Child Abuse Scandal (where social workers were asking children questions like "Tell us the truth, you're not going home until you tell us what Daddy did to you").

The police can, and do, subject people to coercive interrogation techniques. These techniques have been designed to extract a confession, whether you're guilty or not.

Sure, but are that point you are already screwed, and it doesn't matter what you do.

Even if you ARE innocent, it doesn't matter.

The point of security measures is to make you indistinguishable from someone who's innocent.

Yes, you can counter this by saying "well they'll jail you anyway, even if you ARE innocent". But that's a different discussion entirely.

Doesnt seem very different than flushing stuff down the toilet when you see the cops pull up to your door with a search warrent

that sounds like a really good idea and not that difficult to implement.

Truecrypt always had that - one password decrypts into your real partition, the other decrypts the prepared one, with a clear OS. It's impossible to say there is anything in the space where the hidden partition resides, and you have provided a password so at least in theory you are clear.

But it is possible to say that you have Truecrypt installed. Then the LEO just asks for the second password.

> Then the LEO just asks for the second password.

But it's entirely _optional_ to use. They cannot prove there is a second password.

They're not dumb. You installed the thing you're claiming gives you plausible deniability -- it's a fair inference that you wanted just that so that you could hide something, ergo the court will just demand that password.

Which means you shouldn't install any such software unless you're going to have those hidden partitions and are willing to let LE have access.

Do most users of TrueCrypt/VeraCrypt use this "hidden volume" functionality, or do they just want the basic full disk encryption? It wouldn't surprise me if users of plausible deniability were in the minority.

If you could show that most users never do this, you might have an argument in court. How would you do this though?

At the end of the day, the fact that you have this sw installed will make it fair to infer that you are using this feature. You can disagree, but it's what the courts say that matters here, not what I say.

The problem is that you could totally have truecrypt installed and not have a second hidden partition. There is no way to prove that you do. Of course, the court could throw you in jail for not being able to provide something that doesn't exist anyway, but yeah....I'd hope that the lack of proof here would help you win the case.

A prosecutor, a jury, and a judge, will all have a chance to decide that you're telling the truth. There is not mathematical proof here (there never is in any trials, really), so it will all come down to: do we believe the LEO's and the defendant's experts' testimony? Recall, the defendant won't testify -- their lawyer won't let them -- but if they did, they'd probably do themselves in anyways, the court (jury, judge) won't believe them.

There is no plausible deniability here, really.

Putting someone in jail because they won't provide a password to a partition that might exist is truly beyond the pale. If that happens we have completely discarded the notion of presumption of innocence.

You can easily go a 3rd level (more) deep by adding another veracrypt file within the hidden partition. You could also use stenography techniques to hide your encrypted file/data within an image or video.

There are also use-cases that don't include a hidden partition, ex: I keep private code repositories in a veracrypt partition to protect them in the event of laptop theft while not avoiding the perf hit of full disk encryption. There's no need for the hidden partition / plausible deniability for a use-case like that.

Which is why you have 10 passwords, only 1 of which has the important stuff on it.

You are industiguisable from someone who has 9.

What, are the police just going to keep you in jail? What if you really ARE innocent? Jail for life then, because they didn't find anything incriminating?

Free space management needs imply that at some point you must provably have all N passwords or that you're wasting some amount of space just to avoid that, but since that's dumb the fair inference is that you're not doing that, and that you have N passwords to reveal.

Tools like this do not do free space management unless you've unlocked all hidden partitions. You are responsible for that. If you fill your partitions beyond capacity, you corrupt the other data in your other hidden partitions. Of course, unless you've unlocked all the hidden partitions you care about - then it can. However, well designed tools that do this have no way of knowing if you have.

LEO are not stupid. If you are using tech with this feature then they'll ask for both passwords.

N-passwords it is then.

If it's impossible to prove that you've set up the second password, they can't reasonably demand that you provide it if you deny that it exists.

They could unreasonably demand a working second password and incarcerate/torture you until you either provide one or die, but that's the next level.

LEO and courts will (reasonably!) make the following inference:

  "MIGHT be another password-protected partition"
    ->
      "is a password-protected partition"

How does the Truecrypt version of this idea work? I was under the impression that the double-secret data were written in such a way as to be indistinguishable from random bits that normally appear in the files that contain encrypted volumes, with no metadata (e.g. a partition table) to suggest that those random bits are actually interesting.

I figure there is at least one good reason to include amenable "filler" in a "single-secret" file containing encrypted data: it means the size of the plaintext data cannot be inferred from the size of the file. So there is plausible deniability for having amenable "random" data present on that basis alone.

I'm not saying it's easy to extend this scheme to accommodate all environments and/or user experiences. As you say, you can't just e.g. have another partition sitting around full of nothing but suspicious random data.

Some thinking along these lines: https://web.archive.org/web/20100820175505/http://iq.org:80/...

Reply on news.ycombinator.com