Tales from the SOC: Classic Crypto – Operation GUNMAN | S1 Ep023

LISTEN NOW

Classic Crypto – Operation GUNMAN

Cold War cryptographic capers in a real-world Spy vs. Spy scenario!

If the media player above doesn’t work in your browser,
try clicking here to listen in a new browser tab.

LISTEN IN YOUR FAVORITE APP

Find TALES FROM THE SOC on Apple Podcasts, Audible, Spotify, Podbean, or via our RSS feed if you use your own audio app. Or download this episode as an MP3 file and listen offline in any audio or video player.

READ THE TRANSCRIPT

[FX: PHONE DIALS]

[FX: PHONE RINGS, PICKS UP]

ETHEREAL VOICE. Hello, caller.

Get ready for TALES FROM THE SOC.

[FX: DRAMATIC CHORD]

DUCK. Hello, everybody.

Welcome back to TALES FROM THE SOC.

I am Paul Ducklin, joined by David Emerson, CTO and Head of Operations at SolCyber.

Hello, David.

DAVID. Hey there!

DUCK. You have a big grin on your face, because we’ve decided to do something which is educational, but maybe a bit more historically funky than we’ve done in the past.

This is an absolutely fascinating story, with loads of cybersecurity lessons we can still learn from to this day, and that is: Operation GUNMAN.

And maybe kick us off with the precursor to it, which I think goes by the name of the Great Seal Attack.

That’s “seal” as in the Seal of the United States, not “seal,” the aquatic mammal.

DAVID. Oh, not a pinniped?

DUCK. That’s the word I was looking for, pinniped.

But I couldn’t quite find it in time. [LAUGHS]

The coat of arms of the United States of America.

DAVID. All of these stories indicate that there’s nothing new under the sun.

There’s just a million ways to get information from the unsuspecting.

The precursor to GUNMAN was this Great Seal․․․

․․․not the pinniped, but a wooden carved thing that schoolchildren gave the US ambassador to the USSR, and it appeared to contain nothing at all.

DUCK. And this was a time when it was quite reasonable to accept that gift.

The Soviets were allies during the Second World War; the Nazis had just been conquered in Europe.

DAVID. Oh, you practically have to take the thing!

DUCK. Let’s put it on the wall – it’s lovely [LAUGHS].

Big mistake. [LAUGHTER]

DAVID. 1945.

And I think that was the first time that anyone had found a bug that didn’t have a power source.

In this case, the power source was a radiated beam from somewhere else, so a remote power source, essentially.

DUCK. So it’s not emitting any radio waves anyway when you test it and scrutinize it.

DAVID. And embedded in the center, carved in, in a rather intricate way, was this passively powered thing.

When the spies wanted it activated, they would do so remotely using a radio beam.

It was just reflecting sound information back, so it was audio.

DUCK. Right.

DAVID. And so that got you essentially an audio recording of the room.

And, today, we have a version of this too.

Today, if you want to know what’s going on in a skyscraper next to yours, there are such things as laser microphones.

You aim them at the window of a building, and you can hear the conversations occurring in that room, or, in some cases, you can hear someone typing, and maybe define what characters they’re typing.

But all of that is to say – this is still something that we do, this notion of passively measuring at a distance, and powering it at a distance.

DUCK. So it’s basically an unpowered transponder.

DAVID. Yes, it’s an unpowered transponder that, when it receives radio waves at a certain bandwidth that excite it, would then suddenly become a powered transmitter, and would be transmitting back the thing that it was experiencing, which is sound pressure.

DUCK. You have to forgive the Americans for not figuring that out in 1945.

It would have sounded then like the thing of sci-fi.

DAVID. It’s very elegant.

It carries no power source; it basically would look like a wire.

It would look like an RFID tag nowadays, like that.

It’s essentially the same thing that we see today as an RFID tag.

DUCK. Yes, yes, quite right!

It doesn’t do anything, unless it comes into the right sort of environment.

DAVID. It just wouldn’t look like much.

It would look like a disconnected piece of wire, and, in 1945, I don’t imagine that was particularly suspicious.

Not to mention it was hidden in the seal.

DUCK. And if it’s the Great Seal (listeners can’t see me spreading out my arms)․․․

I guess that was to make it look artistically fantastic, but it also meant it’s not going to fit into an X-ray machine very easily. [LAUGHTER]

DAVID. You don’t dig into the Great Seal!

DUCK. Yes!

So when the Americans realized this had been going on, I presume they panicked a little bit and they thought, “Hey, we need to rethink how we do security against eavesdropping.”

DAVID. Now what we knew, essentially, was that you could remotely power something.

The leap beyond that represented the eight or so years in which the KGB was reading the written text coming off of typewriters in the embassy.

These were IBM Selectric typewriters, which, if you know anything about the IBM Selectric, it’s actually a really neat design.

I would particularly recommend you look at the design of the keyboard itself, which is a grid – it’s sort-of an interference grid where every character that’s pressed influences bars in different directions, such that there’s only one combination that can represent that character.

DUCK. It’s like DTMF tones for a modern digital phone.

DAVID. It’s like a gigantic DTMF, yes!

DUCK. And for people who haven’t heard of the Selectric, they were also known, for obvious reasons, as the golf-ball typewriter, weren’t they?

Because the printing ball was a sphere with beautifully formed letters on it.

And they were made in America, right?

They were not gifts from Soviet schoolchildren.

DAVID. They were not!

They were from Lexington, Kentucky, and this represents a supply chain attack.

That revolving ball was cast metal, and it was actually ferrous metal.

And so the exploit that we’re talking about here was a swap of certain components in the Selectric typewriter, one of them being the ball, which went from a ferrous cast metal to a non-ferrous metal with ferrous elements in it, so that you could watch the magnetic field change as the ball moved.

And a bar, a bar placed under the keyboard that would then basically be able to read how that ball was rotating, and encode those rotations into a compressed series of characters that could be transmitted remotely.

DUCK. So that’s how the eavesdropping worked.

It was basically, if you like, some kind of polar coordinates of where the ball had rotated and how much it tilted.

DAVID. Yes, and if you know, or can imagine, what we’re talking about in terms of the design of the Selectric typewriter, what we’re describing would take someone about 20 to 30 minutes to do.

It’s pretty quick.

You put the bar under the keyboard, and you swap the ball out for the ball that is prepared.

DUCK. But the replacement ball doesn’t have any circuitry in it, right?

If you X-ray it, it looks quite normal.

DAVID. Basically, it would be magnetic in a way that the other ball was not.

The shell was ferromagnetic in the original IBM design, but in the Soviet implant, the characters themselves were magnetic, but not the shell.

DUCK. Oh, right, so if the whole thing’s magnetic․․․

DAVID. ․․․You can’t tell where it’s moving.

DUCK. Ahhhh!

“It’s gone around 60 degrees and up 30, so the letter coming next is M.”

If you’re a LinkedIn user and you’re not yet following @SolCyber, do so now to keep up with the delightfully useful Amos The Armadillo’s Almanac series. SolCyber’s lovable mascot Amos provides regular, amusing, and easy-to-digest explanations of cybersecurity jargon, from MiTMs and IDSes to DDoSes and RCEs.

Even if you know all the jargon yourself, Amos will help you explain it to colleagues, friends, and family in an unpretentious, unintimidating way.

DAVID. Yes, exactly.

Exactly!

And so, really, the ball would not tip you off.

What would tip you off would be the bar under the keyboard, which was the encoder.

DUCK. But that bar was a normal reinforcement bar that they just hollowed out, and poked in this circuitry that could only be uncovered by a special kind of X-ray?

DAVID. Right, right.

So the bar itself as a structure was there, but was not filled with circuitry.

DUCK. And if I’m not wrong, David, the other clever thing, at least in the early days, they couldn’t really power them with batteries, because the batteries would run out.

So they actually did a thing like today’s wireless chargers.

They just had some metal near the AC power supply that bled off a little bit of energy to power the hidden circuitry.

DAVID. So, this started in about 1976.

There were multiple generations of this attack that had occurred by the time it was discovered.

None of them tampered with the power supply.

And that was one of the first places that the efforts to determine whether there was a bug would look.

Was the power supply itself a bug, because that would be an easy place to plant a bug?

And all of this was tipped off in 1983, because France discovered a Soviet implant in some of their own equipment at their embassy.

So, the NSA got suspicious, and basically decided that there was no course of action but to swap all of the electronic equipment, all of the Selectric typewriters, out of the embassy.

This was made challenging because it was winter, and they only had access to an outdoor crane that was frozen solid.

So to swap ten tons of electronic equipment without using an elevator, only using an electric crane on the outside, would have to wait till spring.

DUCK. Is it true that the Soviets turned off the elevator probably not as anything to do with operational intelligence, but just out of spite? [LAUGHS]

DAVID. I don’t know if it’s true or not, but that’s obviously the assertion.

DUCK. [LAUGHTER]

DAVID. What that caused, in a practical sense, is a surge of activity in the spring.

DUCK. Right.

DAVID. You could get away with something like swapping all the Selectrics out without the Soviets noticing.

And, really, without your own people noticing.

If you had an insider threat, you wouldn’t want them to know that you’d swapped their typewriters.

DUCK. So I presume that this swapping out tens of tons’ worth of stuff, and including swapping things that didn’t really need swapping, was what’s known today as an “abundance of caution”?

Combined with, “We actually don’t want them to know the particular thing that we’re interested in.”

DAVID. They called it preventative maintenance.

DUCK. lLAUGHTER]

DAVID. They had a bunch of excuses for it.

And in the end, they flew 50 typewriters and a whole bunch of other gear back to Fort Meade, Maryland, which is about maybe 40 miles north of me right now.

They got all of the typewriters there, still not knowing whether they were bugged or not, just knowing that something was up, and they’d found this device in France.

So now they’re back in Fort Meade․․․

They subjected these machines, including the Selectrics, to all kinds of tests.

And they actually used more Polaroid X-ray film than would normally have been produced in three years.

DUCK. Wow! [LAUGHS]

Crikey!

DAVID. And they discovered something – a device that would transmit on frequencies overlapping with Soviet television ranges.

So that’s like the 60MHz, 90MHz ranges.

And also that would only transmit so quickly, so briefly, as to not really be registered on a spectrum analyzer, even if you tuned it to 90MHz and decided that that wasn’t just a television signal.

This was something that someone could install in about 30 minutes.

It was really elegant.

It could be turned off remotely, so if people were getting suspicious, you could just shut the thing off.

DUCK. So, if you’ve got 30 minutes per installation, that means that if the thing is in a bonded warehouse somewhere, even if there’s quite good security, you don’t need a lot of time to do the tampering?

DAVID. You really don’t.

Or you just need a typewriter repair person, or something.

It doesn’t have to be super-sophisticated.

They don’t have to sit around it for five hours, and you’re wondering what that typewriter repair person’s doing.

When they did this swap work, when they basically said, “OK, we’re going to swap out ten tons of typewriters,” they prepared typewriters for the exchange which had been painted in a special kind of paint that was essentially going to seal every component to the board.

They had used staples and fasteners and rivets that had been prepared and engraved in ways that would not be reproducible.

They were transported, including with spare parts, in bags that were not available anywhere in Europe.

DUCK. These were the replacement ones?

DAVID. These were the replacements.

DUCK. So they wished they’d done that before, in other words?

DAVID. Well, they probably wished they’d done it before, but more to the point, they could have done that in the first place and you still would have had the same problem, because they were swapping out two parts.

They were really quite accessible; they were not sophisticated parts.

The presumption was that this was probably some sophisticated component on the main board, or a portion of the power supply.

And really what was going on was two very mundane, structural parts, that get swapped all the time, were being swapped.

The new ones didn’t have the vulnerability, but at the end of the day, this was sophisticated enough in its simplicity that it exploited this assumption gap.

“We assumed that the bugs would be in some esoteric, sophisticated part, deep in the supply chain in Lexington, Kentucky.”

In fact, they were locally swapped, potentially.

The thing that you’re not protecting is the thing that they’re attacking.

DUCK. And probably, if you went, “Well, let’s look for radio emanations,” you wouldn’t expect it to look like something that was mixed in with a local TV signal.

DAVID. Certainly, snuggling those frequencies up makes it easier to evade detection.

Bursting them makes it easier to evade detection.

DUCK. Yes.

DAVID. But there’s nothing to do under the sun.

This is a supply chain attack from 1976.

It’s conceptually identical to what we’re seeing today with TPLink and SuperMicro.

And for all of these companies․․․ it really still isn’t necessarily provable.

TPLink disputes that they are an arm of the Chinese government.

Who knows?

SuperMicro says that they weren’t in on the Bloomberg machines that got shipped.

Who knows?

It’s really difficult to prove any of these things, because supply chains are convoluted.

A lot of people touch your package before you get it.

And that was true in 1976, too.

DUCK. So, David, it’s amazing how long this lasted, which is a story that repeats itself these days with bugs that were introduced – in some cases, not deliberately, entirely by mistake – and just nobody noticed.

This project started in 1976, you say?

DAVID. ’75 or ’76, yes.

DUCK. The French got suspicious in the mid ’80s?

DAVID. In 1983, August ’83, the French were suspicious.

DUCK. And the bugs were only actually found in the returned Selectric typewriters in 1985?

DAVID. Correct.

They were in place until 1984 or so – I think it might have been 1984, late 1984, that they were found.

Really, quite a few years.

DUCK. And all of the things that everyone would have been minded to look out for were deliberately and conspicuously absent.

There was no need for antennas to be buried in the walls of the embassy to pick up the emanations.

Like the Great Seal, this stuff wasn’t necessarily working all the time.

DAVID. It was not.

DUCK. If they’d got wind of the fact that there was some suspicion, they could just lie low for a little while?

DAVID. They found․․․ I think it was 16 devices; 16 Selectrics that were infected, essentially, that were middled.

Those 16 represented multiple generations of this technology.

They all had a memory implant that was the bar, but the way in which that bar was powered changed over the years.

The number of ferromagnetic registers on the Selectric ball changed over the years.

So there was evolution in this.

Which is more surprising, if you think about it – they went on long enough that they were iteratively improving this fundamental design.

DUCK. And also, I guess, all the things that the NSA had thought, “Hey, we really need to protect against․․․”

We need to look for continuous radio emanation; we need to look for specific frequencies that aren’t used for anything else; we need to put lots of filters on the mains power.

And all of that stuff didn’t matter, because that’s not how these implants worked.

DAVID. There was an anecdote, actually, in one of the things I read, that around this time, it was considered unsophisticated that Soviet staff were not allowed to use electric typewriters of Soviet design.

And it appears to have been partly because they knew of this vulnerability; that they didn’t want in on this new world of interceptable typewriters.

DUCK. Ahhhh, so the Soviets were happy to be thought wanting?

DAVID. Yes! [LAUGHS]

The thing that I read indicated that they would only deliver manual typewriters – mechanical typewriters; no power source whatsoever; no kind of fancy grid-bar system – in diplomatic pouches.

And then they would be used, and sent back in a diplomatic pouch.

And so it avoids the customs issue, which the US embassy had, and it also avoids the surface area, really.

The Selectric is a complicated machine, and it isn’t something you just slip in a diplomatic pouch and know that it doesn’t have a bug in it.

DUCK. So the lessons that we can bring forward from this, even though we’re coming ahead 50 years now, are, as you say, “Nothing new under the sun.”

Implants are not always obvious, even if you’ve got a lot of experience in what to look for, based on the past.

This was quite revolutionary, wasn’t it, for its time?

DAVID. It was, and it was an evolution of a generation from – not the pinniped, but the Great Wooden Seal.

DUCK. Yes.

DAVID. Which really is the same cultural design, that comes from the same engineering tradition, but this was the ’70s, ’80s version of that.

Definitely a generation ahead in terms of technology.

When they got all this gear back to Fort Meade, they had a team of people looking at it.

It was all electronic equipment, but a lot of it was cryptographic equipment.

And almost all of the initial attention was paid to the cryptographic equipment.

DUCK. Of course.

DAVID. They were thinking, “Oh, they’re after the cryptographic equipment, that’s what you’d want to be after.”

And they found nothing, and they spent months on this.

And then they took all of these X-rays of just everything.

They burned three years’ worth of Polaroid’s production in X-rays.

DUCK. [LAUGHS] Wow.

DAVID. And eventually turned their attention to the last things in the room, which were this pile of IBM Selectrics.

They X-rayed the whole thing, slice by slice, and the keyboard area on the X-ray gave it away.

This bar that should have been a metal bar was actually still mostly a metal bar, but it was densely packed with circuitry.

DUCK. So even if the X-ray didn’t actually reveal the circuitry the first time they did it, I presume it would have shown slightly incorrect density, at least?

So they’d go, “Let’s cut one open and see what’s inside.”

DAVID. Yes, I mean, it probably was inconsistent with the other ones there.

It wasn’t even most of the Selectrics, it was 16 of them.

DUCK. Ah, right!

So that sort-of suggests it wasn’t a supply chain attack out of the US.

Although I suppose, if you were cautious and you had an insider who would slip things into some Selectrics in Lexington, Kentucky, you might tell them, “Only do one in a hundred.”

DAVID. Yes.

DUCK. Or only do one when you pass a blue car on this corner on the way to work.

DAVID. The only inference you can make, and this inference was made in this case (and it was also made in the SuperMicro case), is that it was probably closer to the delivery destination than Lexington, Kentucky.

DUCK. Right.

DAVID. Not that it’s impossible to have been in Lexington, Kentucky, but the machine wasn’t necessarily being made for the State Department.

It was just a Selectric typewriter in Kentucky.

DUCK. You don’t need to Trojanize all of them as long as you have just enough to get a reliable source of inside information.

End-to-end encryption is not much use if you already know some of the material that’s going to go into the encryption system, is it?

DAVID. Right.

SuperMicro used the same defense in the Bloomberg case.

This is a similar sort of case.

It was a rather quotidian thing to swap.

It was not like they swapped the resistors on the main board, or something that would have been potentially difficult to do and factory-only.

DUCK. Yes, or a completely new power supply with three extra coils in it.

DAVID. No, it’s not that.

It’s like 30 minutes of “new ball; new bar.”

DUCK. Wow.

David, I’m conscious of time.

So, what do you think is the big lesson out of all of this, other than that there is nothing new under the sun?

What do you think is the lesson that everybody can learn, so that they can avoid getting into a situation like this, where you kind of have to assume that everything could be at risk until you’ve analyzed the whole lot, and find out that maybe it was just a tiny part of the whole?

DAVID. Well, I think in a modern sense, or for a modern audience, this is a supply-chain security story that is still relevant.

DUCK. Yes.

DAVID. Where do you get your stuff from?

How do you validate that your stuff is what you thought it was?

That could be true of software; it could be true of hardware.

If you can checksum it, checksum it.

If you can validate a design, do so.

Buy from reputable sources.

Have some assurance of provenance of the equipment that you’re getting.

I would say a modern audience would also react to this being, as I mentioned before, an assumption gap.

We assumed that they would be after voice; we assumed that they would be after cryptography․․․ ․․․but they went after regular old plaintext typewriters.

DUCK. Yes.

DAVID. There are so many ways in which we are at ease with certain forms of communication, or certain forms of expression or storage.

And that’s because we consider them low value.

But those low-value things really might not be so.

And, in fact, they might be the softest target in the room.

The cryptography box is too complicated – don’t bother with that.

Get the Selectric on the other desk!

DUCK. Yes.

Do you think there’s a mirror in the recent Axios story?

“Hey, we could compromise a private key; we could try and crack an encryption algorithm.”

Or we could just call [the chief Axios programmer] up on a Teams call, persuade him that he needs some new codec, and get him to download the malware for himself, by just being persuasive enough.

DAVID. Oh, yes!

DUCK. It’s a very unsophisticated attack, ultimately with exactly the same result.

And, basically you’re getting the other person to do it for you.

DAVID. I’ve mentioned this before on this podcast, I’m sure, but it’s one of my favorite XKCD comics.

The cryptography nerd imagines what the attacker will do to break their algorithm to get at their passkey.

And then the next slide in the comic is the attacker has a $5 pipe wrench, and is just going to beat the daylights out of the person who has the information until they give it to them.

And that is how the world works.

They’re probably not going to attack the extremely hardened point if they can effect the same outcome from a much simpler activity.

DUCK. Yes.

And in a remote case, where you can’t actually get at the person and physically threaten them, you basically flatter them, wheedle them, make them feel comfortable․․․

․․․and then present them with something they need to do where they just don’t stop for long enough and say, “What am I thinking?”

DAVID. Yes, yes, right!

DUCK. Talk someone into tripping over their own feet.

So be very careful of that, folks!

DAVID. I think the last thing, probably, for a modern audience, is we have this burst transmission on TV wavelengths, and those are essentially data exfiltration techniques.

Short transmissions hiding inside legitimate traffic; command-and-control activation.

DUCK. Yes, absolutely.

DAVID. This sort-of just looked like part of the background radiation of the day.

DUCK. Yes, it’s like crooks exfiltrating data with a few DNS requests that are otherwise unexceptionable.

DAVID. Precisely.

DUCK. So, David, I don’t know what to say to conclude, other than, “Be careful out there, folks.”

And remember that the devil can be in the details, and the details can be absolutely anywhere and everywhere, not necessarily the “obvious” places where you might first think to look.

So, David, thanks so much for that fascinating story.

It’s great to have a look back at some Cold War intrigue, but in a way worrying to think, “Golly, we’re still at risk of exactly this same sort of problem.”

So, thanks to everybody who tuned in and listened.

If you like this podcast, please like and share us on social media.

Please subscribe, and please leave us a nice comment if you can – that really helps us a lot.

Don’t forget that there are loads of stories like this, about past, present and future, on solcyber.com/blog.

Until next time, stay secure․․․

DAVID. There you go!

[FX: CALL ENDS]

Catch up now, or subscribe to find out about new episodes as soon as they come out. Find us on Apple Podcasts, Audible, Spotify, Podbean, or via our RSS feed if you use your own audio app.

Why not ask how SolCyber can help you do cybersecurity in the most human-friendly way? Don’t get stuck behind an ever-expanding convoy of security tools that leave you at the whim of policies and procedures that are dictated by the tools, even though they don’t suit your IT team, your colleagues, or your customers!