De-anonymizing ransomware domains on the dark web
blog.talosintelligence.com
blog.talosintelligence.com
I don't know what it is about people who run these criminal enterprises on the darknet, but they constantly seem to be failing even the most basic of opsec. Re-using identities across multiple services, using e-mail addresses with real names, posting photos with identifiable information (and before websites stripped metadata for them, often posted with metadata), etc. I mean it's nice that they are making it easier to catch themselves, but at the same time I can only wonder how some genius can invent some novel and complex ransomware operation just to turn around and use the email they've had since they were 13 to register the services that operate it.
My point is that those mistakes are made by plenty of ransomware gangs, some of the largest dark markets to ever exist (AlphaBay, Silk Road, etc.), Freedom Hosting, and more. All of which were, at some point, major entities on the darknet making absolutely rudimentary opsec mistakes.
Some of the people caught on those listed examples had great Opsec... until that one time where they messed up and then suddenly ended up in jail.
AlphaBay used their regular hotmail account to send password reset emails, and that email was tied to their LinkedIn.
Freedom Hosting was taken down because the operators used outdated FF with javascript enabled.
Silk Road's Ross Ulbricht posted his personal Gmail address, linking the identities.
All of these are profound opsec failures, not just an oopsie that led to getting caught by talented LEOs.
The tightest opsec I've ever seen is maintained by disability fraudsters. Privacy laws protect the evidence anybody would need to present against you, so as long as you keep doctor-hopping and never admit to anything, nobody can touch you. These people tend to be reclusive and not public-facing, but with such low risk comes low reward-- there's no real money to be made in it.
(...unless you're the doctor knowingly signing off on false diagnoses. This increases scale, at which point, the more of those you write, the greater the chances of some mistake made by you or any single one of your patients bringing the whole enterprise down.)
They said that some of the 3 I listed by name had "great opsec". I am curious which one of those they thought was great, and laid out why I think the opsec in these cases was really far from "great".
Maybe when they said "those listed", they were referring to the list on the website and not my list. In that case, I misunderstood and obviously my comment doesn't make much sense. But I presumed they were referring to my list.
>Humans make mistakes and act irrationally at times. Criminal enterprises are complex.
Agreed on both fronts.
But I think that the severity of mistakes is a scale, and some of the really big players on the darknet have made mistakes that I argue is much closer to the "really dumb mistake, trivially avoided" end of the scale, such as using your LinkedIn email to run your multi-million dollar black market.
>Opportunity for mistakes increases with scale.
Agreed. But none of the three examples I listed by name were affected by scale. Using outdated software with known vulnerabilities, posting your own email, and using an email connected to your LinkedIn are all not issues of scale.
Edit to clarify, as I think people may be misunderstanding me (maybe? hard to tell from just downvotes and no replies):
Opsec is hard. 100%. You have to maintain it basically forever, which makes it really hard.
But, if I walk into a bank intending to rob it and start shouting out my full name and address (or, say, left my drivers license at the scene), people would have a jolly laugh at how bad of a robber I was. This is analogous to using the same email to run your multi-million dollar black market as well as sign up for a LinkedIn account. Most people would agree that in my hypothetical, the robber made some really trivial mistakes. I'm not sure why it's so hard to say that for these darknet operators that basically did the same thing, but in computer form.
Dwell time is important too. The longer you stay in the game, the greater chance you’ll slip up.
Remembering that you only have to make an error like that once.
And if all these high-profile people manage to get caught (It seems like pretty much everyone that isn't a nation state ends up getting found eventually!) then maybe it's not that these people are terrible at Opsec, it's maybe that it's much harder than it looks, especially when the government has access to tools that you have no idea about, and maybe it's inevitable that you make an error if you are a human operating for a long time, regardless of 'opsec' skillz.
I tried to make it explicitly clear, over several different comments in this thread, that I'm not saying opsec in general is easy nor am I saying that everyone who has been caught has made these easy-to-avoid mistakes. I am struggling to think of yet another way to word it, but here I go one last time:
A robber goes into a store and steals a bunch of money. On the way out, they leave their drivers license on the counter. Can we agree this would be a dumb mistake? This doesn't mean that all robbers ever caught made dumb mistakes; some robbers are caught through extraordinary police work and with the help of several technologies (DNA, facial rec, whatever). Those robbers, while still potentially making mistakes that lead to their arrest, have not made extraordinarly dumb mistakes like leaving government-issued identification at the scene of the crime.
This concept applies to opsec and computers as well. You can slip up once and be caught through the smallest of mistakes. Or you can literally tell everyone who you are and be caught that way. Both are mistakes, but one is a trivially avoided stupid mistake, and the other is not.
Many other operators (of dark markets, ransomware gangs, etc.) have been caught, but I did not include them because the ways they were caught did not appear to be through dumb mistakes, but through intense technical means.
the server was never seized and the operator was never sei
what about the currently existing markets still up? just one trip to dark.fail to check
It's not a particularly high bar, but I suspect the majority of technically apt people would fail it.
People with lucrative work available legally have more to lose and less desperation to engage in activities that are both illegal and malicious
Being in 85th-99th is no longer attractive. Because of extreme level of taxation people who earn real money are not included in those statistics, as in they are not paid a salary.
In my country (UK) being in 85th-99th percentile, means you'll have okay-ish life. After years of saving you may be able to afford an old terraced house, second hand, couple of years old car. Oh and if you decide to have family and your spouse won't be working, then you are screwed.
These indicators are no longer what they used to be.
as unfair as it may be, a huge part of the usefulness of information is its accessibility, and these search engines currently hold a near-monopoly on which sites can generally be considered readily accessible, ie the 'surface web' above the deep web
That's a really odd way of naming thing.
They are not web, and "not indexed" usually is referred to as "deep web", not "dark web".
They key is to ensure only legal stuff goes out on your IP and the illegal stuff is anonymised. Which is easier said than done.
you can do the same trick to connect to it from home - VPN use is common. you'd want a burner laptop, of course, and some physical box preventing the laptop from hitting anything other than the VPN.
I've thought about setting this kind of thing up for fun. you could get really fancy - talking to some hopbox through Tor where you script up actions to take asynchronously, to defeat timing attacks.
Same problem. Tunnelling all of your traffic will look suspicious and thus stand out from the thousands of other people who don't tunnel all of their traffic. If I recall correctly, one of the documents Snowden released even specified that people who tunnel all of their traffic via VPC land themselves on government lists for closer monitoring. Regardless of whether this is true or not, creating a lot of legitimate traffic on your same gateway should still make it harder to fingerprint you as someone who exercises in activities that warrant closer inspection.
This keeps things simple (conceptually) while also effortlessly creates genuinely normal looking traffic. However eventually you'll still get caught. It doesn't matter how careful you are, you only need to slip up once.
When running a darknet site you don't want associated with the clearnet, step one should be only having the http server listen on the Tor onion domain!
https://www.spiceworks.com/it-security/cyber-risk-management...
20 year old memories of proxying my ssh traffic through InterNIC just came flooding back!
SSL may stop your roommate or isp but they provide another vector for linking to other entities.
I wonder how many are using this technique to link web properties together.
Right now, SSL(or PKI to be precise) is a very privacy respecting technology. For both the server and the client.
Anyway, these all seem like pretty obvious opsec fails where the darknet website is also served over the regular internet, which is just atrocious.
Good advice
>they do provide privacy against snooping exit node
onion services don't use exit nodes. Your client and the service build circuits to nominated middle relays so https only offers very marginal increases in privacy. However, you are right to assume than any exit node may (or probably is) monitored.
Thanks for the clarification.
For end-users TLS and Tor both provide privacy; since you don't need to identify yourself in order to use https. In fact, with ESNI and DoH the only thing anyone snooping wire traffic can see is that you're connecting to whatever data center is owned by the company hosting the website.
The sites in the original article are criminal enterprises, which means they have the unique problem of needing the origin server to remain anonymous so that their hosting provider can't find out what they are doing. This is the one thing Tor does that TLS doesn't; and they were deanonymized by them insisting on providing a self-signed cert anyway. However, this is a particularly unusual threat model that is far harder to maintain. Even the whole anticensorship thing is usually just hiding what sites you're visiting from, say, the Great Firewall - we don't care that China can also use Tor to learn where Google's servers are.
Granted, that's still kind of pointless because you still have to self-sign, which gives scarier warnings than being unencrypted[0].
A knowledgeable user wouldn't care - they'd know that they installed a Tor gateway that resolves .onion to itself, so they're just as protected as they are on TLS. The catch here is that the ransomware operators are trying to criminally extort less-knowledgeable users and bureaucratic IT staff that are just being told to "run Tor and pay us in Monero to get your files back".
[0] There's nothing preventing these operations from shipping their own browser or root cert - they are, after all, already running on the local machine outside of any sandboxing. No clue if they do this.
Generally, though, TLS is not designed with privacy of the server in mind. The data exchanged between the client and the server is kept private between the two parties, but that's it.
If you wish to anonymise your connection, technologies like Tor will help. You'll still have to pay attention though. In a great many cases, security and usability are polar opposites, and a balance must be struck to find a workable solution. In this case the best balance is probably in-depth knowledge of how web servers work combined with reading through the documentation of the Tor project.