USB Drives and Wax Seals

Need some pre-industrial security for your USB drive? How about a wax seal? Neat, but I recommend combining it with encryption for even more security!

Posted on May 16, 2012 at 1:50 PM • 38 Comments

Comments

DanMay 16, 2012 2:15 PM

would be neat...except that it comes sealed, so you have to break the seal to put any data on it rendering the whole thing kind of pointless.

The "DIY" version at least lets you make your own seal after you put some data on it, which could make for an interesting conversation piece.

stvsMay 16, 2012 2:18 PM

DRUMMOND: “If I had Golden Dancer I’d have everything in the world that I wanted.” (He cocks an eyebrow) I was seven years old, and a very find judge of rocking horses. (He looks off again into the distance) Golden Dancer had a bright red mane, blue eyes, and she was gold all over, with purple spots. When the sun hit her stirrups, she was a dazzling sight to see. But she was a week’s wages for my father. So Golden Dancer and I always had a plate glass window between us. (Reaching back for the memory) But—let’s see, it wasn’t Christmas; it must’ve been my birthday—I woke up in the morning and there was Golden Dancer at the foot of my bed! Ma had skimped on the groceries, and my father’d worked nights for a month. (Re-living the moment) I jumped into the saddle and started to rock— (Almost a whisper) And it broke! It split in two! The wood was rotten, the whole thing was put together with spit and sealing wax! All shine, and no substance! (Turning to CATES) Bert, whenever you see something bright, shining, perfect-seeming—all gold, with purple spots—look behind the paint! And if it’s a lie—show it up for what it really is!

Jeffrey GoldbergMay 16, 2012 2:25 PM

I loved playing with sealing wax as a kid. Mostly, though, it was just an excuse for playing with fire.

Although I can't imagine a situation where this would be of any real use, given the ability of the attacker to reseal the device, I still want to do this. It's sort of steampunk or something.

Alan BostickMay 16, 2012 2:34 PM

Isn't it the case that long before strong crypto came about, various Black Chambers were able to open and reseal wax seals virtually undetectably?

aphMay 16, 2012 2:35 PM

@Jeffery,

If you mix two colors of wax and take a picture (saving the picture in an immutable location), you can create a seal that is tamper evident and can't be easily replaced.

NobodySpecialMay 16, 2012 3:02 PM

If you want to do it yourself you can get colored sticks for glue guns.

Dirk PraetMay 16, 2012 3:24 PM

Neat gizmo when used as a hand-out at security conferences by businesses wanting to profile themselves in an original way.

scorcheMay 16, 2012 3:29 PM

(Note: I have won the Tamper-Evident contest at DEFCON and help to run other contests at other cons)

@aph - You have the right idea here - this certainly is one step in making things harder to tamper with. Another recommendation is to press the stamp quite hard - a thin seal is a good seal.

@NobodySpecial - you will want to make sure it is a good type of sealing wax. Not just any wax would work for sealing. There are specific properties that increase the level of resistance to attacks.

Yoseph AsherMay 16, 2012 3:59 PM

"Additionally you can select your own classic letter to be pressed on the wax seal."

Ok so you find out what seal he is using then just order one... not really so useful these days of mass production..

vwmMay 16, 2012 4:15 PM

The problem with wax seals is that you can derive a seal-stamp (=private key) from a seal (=public key) almost as easily as the other way around. Aph's idea changes that property in a very nice way.

Now we only need to transmit that picture in a tamper proof way. We can probably solve that issue using recursion ;-)

scorcheMay 16, 2012 4:27 PM

@vwm - Yeah. The idea of creating a unique marbled wax seal has been around just about as long as wax seals have been around.

@waxonymous - No, it really doesn't. As I mentioned before, just "wax" isn't necessarily optimal either.

WimWMay 16, 2012 5:36 PM

@waxonymous chewing gum would be a good idea. Analyzing DNA would be the private key!

SteveSMay 16, 2012 6:47 PM

If it came with a signet ring bearing my family coat of arms, that would be cool.

WaelMay 16, 2012 6:53 PM

Pretty weak.
1- Does not gurantee confidentiality
2- Promises detectability of tampering, which is also weak. It can be bypassed in many ways, including opening the USB enclosure without breaking the seal. Just use a sharp tool, cut the enclosure and reseal it, then touch it up.

Clive RobinsonMay 16, 2012 6:58 PM

@ aph,

If you mix two colors of wax and take a picture(saving the picture in an immutable location), you can create a seal that is tamper evident and can' be easily replaced.

Actually the modern version of this is done with printing a unique serial number on the object using "etching ink" then a layer of clear epoxy resin with a layer of very loose "chop strand mat" fiberglass and on top of this more clear epoxy with a small dusting of small glass beads with a slight stipiling in the finish before the expoxy cures.

You then take the equivelant of a holographic image of the seal.

I've been told that this was considered suffiiciently tamper proof for use in SALT treaties.

In effect the US put one on every Russian nuke or armament that was to be limited. And the Russians would put one on every US nuke etc.

Independent monitoring teams would then make random visits to check the serial number seals.

@ Alan Bostick,

Isn't it the case that long before strong crypto came about, various Black Chambers were able to open and reseal wax seals virtually undetectably.

Yes and No...

There are two types of documents "sent under seal". The first is where a single sheet of paper/parchmment is either written on or wrapped around another message, the second is to use an envelope and seal the flap down.

The second is compleatly insecure and the seal needs not be touched in any way what so ever. Basicaly what you do is bank the envelope on a surface to make the contents go to the top of the envelope. You then clamp the bottom of the envelope between two steel rulers with just a paper thickness of the envelope protrouding through. You then using a very very sharp razor slice this off and take the letter out of the envolope and copy it before returning it. You then re clamp the envelope but this time with just a half millimeter of the envelop protruding you then fold it back very hard fold it back up. apply a very thin bead of glue down the outside of both sides. Wait untill nearly dry then remove from the clamp fold the flaps into the envelope and re clamp and wait untill the glue has set, then put it back in the post. Few people will spot this as they just don't inspect the eenvelopes carefully.

A second and more modern way due to the design of the ordinary white windowed envelope is to go for the glued side seam. If you have a knife with a thin flexible blade that is not to sharp you can insert this at the right angle and using a sawing motion actually cut the glue not the paper. You then unfold the seam extract the letter copy and put it back. You then very gently apply PVA wood glue over the glued area you have cut ensuring the coverage is the same. Then when tacky put the flap back in and apply preasure untill the glue is dry. Even on examining the envelope this is very difficult to detect. It is why sometimes when asked to give important refrences you are asked to not only put sellotape across the envelope flap, but also around the edges and sign over both pieces.

When using seals on the first type of package using a single piece of paper it is best first to squigle in a series of random designs etc or have it printed in raised ruled ink designs (these are used on some bank notes and when you tip them through the light the raised ink throws shadows to form particular patterns). From an attackers point of view cutting this paper will be very tamper evident so they have to go for the seal.

Now unless a particular type of sealing wax is used you can attack the seal by very slowly heating the seal from the other side of the paper whilst keeping the top of the seal as cold as possible (wet cotton wool with an ice cube in a thin plastic "sadwich bag" on top is one way). The trick is to get the bottom of the seal to the "plastic" state between solid and liquid then using a small amount of bending and a very thin slide of glass the top of which has a lubricating solvent on you can slice the seal off.

The way to stop this is to fold the paper in such a way that all four edges come together over seam folded with a rose fold in the middle on which the seal is placed the sealing wax compleatly envelops the rose fold and alows a very small amount of the sealing wax to go through the center of the rose fold onto the letter below.

As a general rule for seals you need to make their recesses deep and with a cross ground surface that produces a secondary image. When you apply the seal to the wax you press it sufficiently hard that it actually displaces the wax away from the paper surface. You thus have the cross cut image showing up from the paper surface, and the main seal image on top of this.

Another way is to use an embossing stamp (like those used by noteries etc) on the edges of the paper then apply the sealing wax over this and press the seal down with moderate force so as not to flatten out the embossing. On opening the paper wrapper you inspect the back of the seal to check that the embosing is clear and undisturbed.

Then there are older methods used with parchment where it is sewn with a thread that is then sealed at the ends. The idea being that the thread should be uncut on delivery and cracking the seal to remove the ends of the thread should enable the integrity of the seal to be checked.

Some modern sealing methods work this way using very fine enamaled copper wire the circuit can be checked to see if it is still integral. Often this is used inside computing "security modules" that are then filled with epoxy with quartz sand in it. As the ends are inside the module along with the battery back up trying to get into the module will break the wire giving the CPU time to erase and start over writting the secret stored in the ram. This sort of thing is by it's self insufficient as there are ways to attack the epoxy without drilling or grinding and the circuit can be frozen in time using various cryo techniques that will stop the CPU but retain the memory contents. There are both hardware and software solutions to both of these attacks that I've outlined in the past (search on "data shadows" my name and "RobertT" or "Nick P").

There is a problem with using threads or cords especialy those on electricity meters where the thread is a piece of stainless steel wire and the seal a soft metal such as lead or whites metal. It is possible to heat the wire with a high power soldering iron and slip it out of the seal. The industry first solution was to deform the ends of the wire in some way so that it was larger in diameter, one such way being to put eyelets or crimps on the end. The method of attack then becomes heat the wire slide it back towards the seal untill the eylet or crimp is against the seal. cut the wire on the other side of the seal as close to the seal as possible. Putting the seal back can be done in a number of ways but basicaly you end up pushing the cut end back into the seal and when it has cooled the seal metal has melted into the wire and wicked sufficiently that the end is effectivly soldered in place. A simpler way if you are any good at "tool making" is to make a new "security crimp" as the whole thing is generaly a deterant not a proper security seal the design of the crimp embossing is generally very simple and not very acurate in design, likewise making a new "unused" seal is usually just a case of identifing the metal its made of sourcing a rod of it and turning it down on a watchmakers lathe (if you don't have one then a 12volt craft drill with a large chuck held in a vice will do the same only instead of using cutting tools you use needle files and grit paper to provide the finish). I know from experimentation and "proving the point" to others you can do the whole lot from scratch in a day...

he23May 16, 2012 7:02 PM

Wax seals is like a certificate chain, where certificate is realized by escutcheon (emblem) and the certificate chain by genealogical tree: descend tree.

What's about prooving identity by insignia or sword in the stone?

Not such classical but more indigenic is IPv4 over smoke signs with pattern recognization!

scorcheMay 16, 2012 7:42 PM

@Wael - Tamper evident devices do not have "guaranteeing confidentiality" as one of their priorities. There are often (or should be) other controls for that. The purpose of a tamper evident device is to alert some party that there may be a breach in confidentiality or integrity.

WaelMay 16, 2012 8:52 PM

@Scorche - You are correct in the pure sense. But If a device is tampered with, and you fail to detect the tampering, what can you say about the contents? Man! the web page says "Top secret" that does not mean "integrity", that means "confidentiality", correct?

Jeff ZimmermanMay 16, 2012 9:04 PM

@Clive Robinson - the holographic "signature" for SALT verification sounds fascinating. My google-fu is weak - do you have any links that further discuss how this works?

WaelMay 16, 2012 9:49 PM

Here is another way to bypass it: poke a small hole facing the connector end, insert your thin USB connector wires, do your thing, then seal the tiny hole :)

WaelMay 16, 2012 10:23 PM

Read the link again... The package is made out of porcelain, so it's not trivial to cut and reseal. Poking a hole maybe easier in the hands of a class 2 adversary. But anyway, this is not a device that will keep your Data "secret" as they claim, unless I misunderstood somthing.

PaeniteoMay 17, 2012 1:48 AM

@Yoseph Asher: "Ok so you find out what seal he is using then just order one..."

There are other companies who sell seals based on an image file supplied by you. You are not limited to the 27 seals offered directly in combination with the device.

Ahead of CurveMay 17, 2012 9:17 AM

Aspect Bell, Quantum entanglement, Spooky Action at a Distance:
Imbed a particle who's spin changes when released. Keep a twin particle. When seal is broken twin particle is changed as well as the original generating an alarm indicating time of receipt or compromise.

epic failMay 17, 2012 10:39 AM

@scorche/datagram

paper has tooth, a rough porous surface.
the usb drive is a smooth piece of plastic
put that in a zero degree freezer for a few hours and take it out and just wiggle the two halves and the connection to the plastic on one side or both may separate so you can remove the cover.

scorcheMay 17, 2012 11:43 AM

@Wael - Which is exactly the reason that so much effort needs to go into training seal inspectors. An "arm's length" inspection only goes so far. Any tamper evident manner is useless without proper inspection efforts.

@epic fail - I am scorche, not datagram. Datagram is another person on the MFP team. ;)
My comments have all really been about wax seals in general - not really this specific implementation. You are totally correct though. There are a variety of different methods to separate the wax from the smooth surface of the USB drive. Wax seals don't work so well on non-porous surfaces...

Surreptitious EvilMay 18, 2012 8:49 AM

When sealing particularly sensitive protectively marked documents (often duplicate safe combinations) which were then subject to long term storage but you needed to know if they had been opened by somebody with access to that storage, the usual procedure was:

1. Seal envelope in normal way.
2. Apply 3M / Scotch "easy tear" tape (special versions of this stuff).
3. Sign in biro and date along each envelope seal.

The assumption was that the attacker had legitimate access to the storage location, plenty of time (a military overnight duty, say) and ready access to the necessary materials to attempt a duplicate.

On the other hand, I used to have a lead seal stamp kit for 'securing' disk and tape boxes. That never struck me as particularly robust.

scorcheMay 18, 2012 11:14 AM

@Surreptitious Evil - The adhesive tapes are one of the easiest things to get past... It is all in knowing the correct solvent to use.

askme233May 18, 2012 2:10 PM

Old post, but I am just catching up:

@Clive on ways to read a letter:

I used to live in Beijing around the time of the democracy movement and all of our mail was always clearly read, despite the envelope seal being untampered with.

On most envelopes the glue does not go all the way up to the top of the flap, but leaves about a quarter inch near the top of the envelope unglued. What we found out they do is to slip a split chopstick underneath the flap at the very top such that it would catch the letter inside. spinning the chopstick rolls the letter around it and allows it to be extracted through the unglued gap. It is returned the same way.

Never figured out why they would go to so much trouble to be "undetected" and then use a black marker to censor certain terms in the letter.

Daniel JanzonMay 21, 2012 3:48 PM

Although I wouldn't trust this seal, theoretically it would be a nice way of sending a one time pad to a friend.

Davd FMay 23, 2012 11:41 AM

I'm really enjoying this chain of comments much more than the one for the Rules for Radicals.

Lev AbalkinMay 27, 2012 5:26 AM

I'm tempted to be facetious, and ask if wax walruses are lacking somewhat ...

Seriously, one of the best stories I've read that involved a marker to indicate an unwanted intruder's presence involved a young man stretching a thread from one side of his room's door to the other. Then he broke it.

He found it stretched from one side to the other when he returned.

When it's guaranteed that someone's going to break the seal, it might be an idea to use that to defeat their intentions.

Leave a comment

Allowed HTML: <a href="URL"> • <em> <cite> <i> • <strong> <b> • <sub> <sup> • <ul> <ol> <li> • <blockquote> <pre>

Photo of Bruce Schneier by Per Ervland.

Schneier on Security is a personal website. Opinions expressed are not necessarily those of Co3 Systems, Inc..