DSS 2026 / Day 1 / Amber Hall
What the Block Explorer Misses
Julia Hardy
Source: YouTube automatic caption track · Exported 2026-11-16. Timestamps link to the source video. Cue end times were not supplied.
Transcript
0:00Okay. Thank you everyone. Um, I'm Julia,
0:03here with Zero Shadow, and I'm going to
0:05be talking about yeah, money laundering
0:08insights and how we can use that to
0:10contain active exploits.
0:22Um, okay. So, um, when a exploit
0:25happens, I think a lot of times people
0:27will hopefully look at the attack
0:29vector, um, and and try to learn from
0:32that, but really focus on the very first
0:34address where funds might have been um,
0:38deposited into, and then that's the end
0:40of their journey. Um, maybe for some of
0:42the larger exploits, some people will,
0:45for example, follow movements to Bitcoin
0:48for across storage chain for something
0:50like the Bybit exploit. But um, what's
0:53unique about Zero Shadow is that we are
0:54looking at every single address that a
0:57threat actor uses from the very first
1:00deposit, uh, the very first stolen funds
1:02address all the way through to the last
1:04address where we see a suspected cash
1:07out.
1:08And that means that we have a really
1:10holistic view of how certain threat
1:13actors are moving their stolen funds.
1:19And the the insight that I get from a
1:21lot of people when I talk about freezing
1:24and recovery, um, along with security is
1:28why should we be focusing on the
1:29laundering side? By then it's too late.
1:32Uh, we should really be just doing
1:33preventative work to try to stop any
1:36kind of um, incidents from occurring.
1:40And while that is definitely important,
1:43I think that the first or the year so
1:46far has really shown us that it can be a
1:49matter of when and not if for certain
1:52protocols to to be hacked, and that
1:55knowing who your adversary is can be
1:57really important to um, actually have a
2:00containment when that exploit occurs. Um
2:03when we look at the first half of the
2:04year, we had
2:05calculated 1.8 billion dollars stolen up
2:09until October and then of course the
2:11past weekend has happened and that
2:13number is now up quite a bit.
2:16Um we've also seen that a lot of these
2:18exploits are happening because of
2:20operational security. So off-chain um
2:23even maybe dependency related if you
2:25think about things like the Axie client
2:28exploit earlier or vulnerability earlier
2:30this year, some of the different wallet
2:32exploits that have been happening where
2:34it's really not just contained to your
2:36specific protocol.
2:40And Zero Shadow launched a new product
2:43this year, the Threat Intel platform. We
2:45have basically taken all of our stolen
2:47funds addresses and we're trying to
2:49better deliver them to the services that
2:51can block funds.
2:53Um and what that's created is a really
2:55big data set of all these different
2:58hacks that we look at, all the different
3:00threat actors that we look at, every
3:02single address associated with those
3:04movements, and we can run analytics on
3:07those um addresses and start to find
3:09some really unique insights.
3:12So what does the data show us?
3:15Um I looked at a few different groups
3:16here. The first one, Trader Trader from
3:19North Korea. This is the really big most
3:22sophisticated exploits. This is most
3:25recently Binance um across a number of
3:27different exploits. And um this this
3:30first stat here is how quickly the
3:33initial funds are actually moved out of
3:36the first address where they are taken.
3:39And this is not swaps or anything. This
3:41is just um actually sending emptying out
3:44that wallet and sending the funds
3:46forward. And you can see that the
3:48ability to react to those initial funds
3:50is
3:51less than 30 minutes. Um a little bit
3:54longer for some of the Bitcoin side, but
3:57um but generally you're really just
3:59looking at a a a very fast emptying of
4:03those wallets, moving funds forward to
4:05new addresses. And so, if you're really
4:07just focusing on that initial
4:09containment, you have a very small
4:10window to try to act.
4:13But what's interesting is that we see
4:15that there's a longer window when the
4:18funds have been maybe spread out across
4:2020 addresses. We're calling these the
4:22staging wallets before they actually
4:25start to do the rapid fire laundering.
4:28Um and why does this occur? This occurs
4:31because North Korea is actually handing
4:33the keys over to Chinese money
4:34launderers to to continue with the
4:37laundering. So this is them preparing,
4:40coordinating with those launderers,
4:41making sure that they can actually
4:43continue with this activity.
4:45Um and we're seeing that this time of
4:47staging is decreasing over time. So it
4:49used to be it used to be more like 6
4:52months. Then you look at last year, more
4:55like 6 days of this pause between the
4:58initial theft and the rapid fire
4:59laundering.
5:01Um now we're looking at more like 3
5:02days, even smaller. So
5:06why is this window important? Most
5:08important, I think, for um anyone who
5:11experiences an exploit to realize I have
5:14only a few days to make sure I have law
5:15enforcement prepared. I have all the
5:18different services that I want to um
5:20know about these addresses prepared,
5:22maybe a security firm, any of those kind
5:25of steps. This is your big window before
5:27the rapid laundering to do so.
5:31And then when we look at just the
5:33downstream laundering, so the the
5:36movement towards the cash out,
5:38most addresses are only holding a
5:40balance for about 5 minutes. And so um
5:43in in Bybit, for example, we saw 13,000
5:46addresses used. And so again, this is
5:49just this idea that you really need the
5:51real-time information if you want to try
5:53to
5:54intercept these funds at any time.
5:58Um and we were actually able to do that.
6:01Um so this is from the Caltal KelpTel
6:05exploit that we saw funds were
6:07originally going to Thorchain and
6:10then the launderers were looking for
6:13more rapid bridging opportunity, so they
6:15switched to other bridges that were
6:18ingesting the addresses that we were
6:20sending to them and we were able to
6:22intercept $700,000 worth of funds. And
6:25this graph here on the right-hand side,
6:28this is all the different services that
6:30were getting hits off of these
6:32addresses. So the launderers trying to
6:35repeatedly send these funds in and
6:37getting rejected and trying again.
6:42And so shifting gears here to other
6:44North Korea groups. This on the screen
6:47here is the Drift exploit. So the the
6:50specific actors here is Apple Juice.
6:54But who are this actor if you're not as
6:56familiar?
6:57Basically these are not as obvious hacks
7:01from
7:03the the specific vector that are used,
7:05but they're always using the same
7:07malware.
7:08Or if you've heard of something like
7:10Dangerous Password where this is the
7:12Zoom call where the audio doesn't work,
7:14these are not as sophisticated of actors
7:17within North Korea, but they're still
7:19doing pretty complex exploits.
7:25So when we look at the fund movements
7:28for laundering for Apple Juice, we see
7:30that the initial theft addresses are
7:33emptied pretty quickly, not quite as
7:35quickly for from the trader-trader side,
7:37but still
7:39between 30 minutes to 3 hours those
7:42funds are really emptied out. And again,
7:44you really need to be thinking more
7:45downstream if you're thinking about
7:47disruption.
7:50But what's interesting here is that we
7:52see that the funds will be parked for a
7:54really long time just sitting at
7:57externally owned accounts before they
7:59move to the next stage of laundering.
8:02So for two different cases here I
8:03anonymized them, but we're looking at
8:0610 to 11 months, 4 to seven months where
8:09funds are just sitting in in wallets.
8:10They haven't been moved.
8:12The reason that we see this
8:15is is partially a reevaluation of
8:18strategy.
8:19And that's really allowed us to have a
8:21lot of time to understand exactly what
8:24we need to be prepared for. So where do
8:26we expect these funds to move to? What
8:28mixers? What lending protocols? Get
8:31everything in order so that you can have
8:33really good results with disruption when
8:35that the time comes for those movements.
8:40Also we are able to see with the data
8:42that 86 to 91% of the time that the
8:45funds are being moved is between 9:00
8:48a.m. to 6:00 p.m. Korea time. And so
8:50unlike with the trader trader funds
8:53where we see pretty much 24/7 movement
8:55and that's very much being automated
8:58with launders. This is North Koreans
9:01moving their own funds, working a
9:03working day and then taking a break to
9:05go to sleep.
9:07And so it's a very different style of
9:08laundering and again gives you a little
9:10more time to to act.
9:15And then just looking a slightly
9:17different type of actor, the non-nation
9:20state actors.
9:22We took a look at
9:24a couple of thefts that we've been
9:25seeing from the com. If you've heard of
9:29the com before, you've probably heard
9:31about these teenagers, early 20s kids
9:35who are a lot of times
9:37calling you up pretending to be
9:40Coinbase security and you need to move
9:43your funds to secure them and you lose
9:45all your funds. We're actually seeing
9:46that that's not the only kind of attack
9:49that they do.
9:51Something like drainer as a service has
9:53definitely been a past
9:56exploit vector, but we're now seeing
9:58actually just taking advantage of of any
10:01kind of vulnerability that they can see
10:03and even doing some more smart contract
10:06based exploits.
10:08And when we looked at a few different
10:10cases that we know are the same group of
10:14actors within the com, what's
10:16interesting with the data is that
10:18the majority of the funds are ending up
10:21going to bridges as one of the first
10:23destinations
10:25as a laundering path. And so this is a
10:28really key area to concentrate on as a
10:31victim of of a skid theft is to make
10:35sure that the bridges know that these
10:37are the potential addresses of source of
10:41funds that you need to be blocking and
10:43freezing if you can intercept them.
10:50Also as another
10:52non-nation state actor, there's been a
10:54lot more rent attacks this year, a lot
10:56more in the news of the UK best based
10:59rent attacks.
11:00With these actors, we see that the
11:03movement of funds can be 4 hours about
11:06between each address. So again, a much
11:08slower rate of laundering. And there's
11:10also heavy co-mingling across victims,
11:12which means the same address is being
11:14used for multiple victims over and over.
11:16So if that address is on a block list
11:19one time, it might have a good result in
11:22the future for another victim. And so
11:23that's even more opportunity to try to
11:26disrupt.
11:31And we did have a recent success with
11:34disrupting a rent attack.
11:36So the victim had posted they had lost
11:39$800,000
11:41on Twitter. Within 30 minutes, those
11:44funds had gone through the threat intel
11:47and been distributed out and ultimately
11:50about 90% of those funds were frozen.
11:52And so there's a lot more of that
11:54coordination
11:56possible, not just with this actor, but
11:58in general by being really quick with
12:01the delivery of these addresses.
12:10And and is hopefully uh the takeaway
12:12that you got here today is that data can
12:14really help contain your exploit. The
12:16common denominator across all these
12:18different actors is that time is of the
12:20essence. And if you're thinking about
12:22how do I prepare for my exploit, the
12:25common questions that I hear time and
12:27time again are how do I make sure that
12:28services know about my stolen funds? How
12:31do I move forward? What are my next
12:33steps? What are my recovery chances? And
12:36I think the data is really the common
12:38denominator here to make sure that
12:40that's possible.
12:43So thank you so much. Here are the
12:45socials for myself and Zero Shadow if
12:47you want to
12:48um follow us and uh yeah, thank you.
Source and accuracy details
- Names, organizations, technical terms, code and numbers may be misrecognized.
- Source wording and disfluencies are preserved; no editorial corrections were applied.
- Export contains whole-second cue starts only; cue ends and subsecond timing are unavailable.
- A full caption-track export does not establish that every spoken word is represented.
- Automatic YouTube captions; not human-verified against the audio.
Source export SHA256: 87413dbed1c77599db1442f6c1a3a08fd7eb3f59b76b63f2e44d54b4f9713943