- Home >
- Resources >
- SecureTalk >
- "It's a Piece of Junk": A Cybersecurity Professor's Case Against Tor
"It's a Piece of Junk": A Cybersecurity Professor's Case Against Tor
One of the most respected voices in online child safety research just called Tor "a piece of junk" — and he has the math to back it up. Brian Levine — Distinguished Professor at UMass Amherst, ACM Fellow, and Director of the UMass Cybersecurity Institute — joins Justin Beals to explain why two decades of fighting online child exploitation convinced him that one of tech's most trusted anonymity tools does more harm than good.
Chapters
00:00 Levine challenges common claims about Tor
02:13 The peer-to-peer research that revealed online abuse
05:36 Rescue Lab and safety by design
07:41 Investigating with public data, lawfully
10:40 Why university research helps - and its funding limits
14:30 Peer-to-peer networks, moderation, and accountability
16:27 Tor Browser versus Tor Onion Services
18:53 Levine’s critique of Tor’s protections and governance
22:38 How online platforms are exploited
25:50 Who controls Tor, and why terminology matters
31:24 Privacy for whistleblowers and dissidents
36:18 AI-generated abuse and designing safer platforms
📝 Summary
Levine's core claim lands hard for a security audience: cryptography isn't the weak point — governance is. "You don't break the encryption of Tor. You disrupt its routing through the network. And it's a very weak system," he explains, arguing the Tor Project has known about documented abuse for years without meaningfully addressing it.
For CISOs and compliance leaders, the takeaway extends beyond Tor: any platform — anonymity tool, marketplace, or SaaS product — has to be designed assuming bad actors will exploit it. "Safety by design," Levine says, isn't optional anymore.
🔔 Strike Graph builds the compliance infrastructure that turns "safety by design" into practice for defense-industrial-base companies. (www.strikegraph.com)
About the guest: Brian Levine holds a PhD in Computer Engineering (UC Santa Cruz, 1999) and directs the UMass Cybersecurity Institute. Elected ACM Fellow in 2020; sole author of a 2022 NIJ Report to Congress on internet-based crimes against children. Featured in the New York Times, Wall Street Journal, and other major outlets.
Related episodes:
CMMC Phase 2 suspension: https://youtu.be/V-eiHMJiYbs?si=L82UtGOJSTFynXQG
AI governance with Fran Berman: https://youtu.be/2_iRwe_AUp0?si=hdFv85muCRWQcxNP
Non-human identity with Dan Cinnamon: https://youtu.be/UQKYuKpE6lo?si=Fkvsr8A1mE2GX24S
#CyberSecurity #Compliance #TorNetwork #ChildSafety #AIThreats #DigitalForensics #OnlineSafety #SecureTalk
View full transcript
Justin Beals: Hello everyone, and welcome to Secure Talk. I'm Justin Beals. Most of us have been told that Tor makes you anonymous and that the dark web is something we can't see into. My guest today says both ideas are mostly marketing. Now, Brian Levine has spent 20 years studying how these networks really operate. He'll explain why Tor Onion services have become the backbone of modern ransomware and why the content people assume is hidden.
Could be found if anyone bothered to look. Now, his bigger point lands straight in our world. The safety-by-design thinking we use on our own systems belongs in every platform before it ever ships. Now, Brian is a distinguished professor at UMass Amherst and directs the UMass Cybersecurity Institute. He's an ACM fellow, and in 2022, he authored a report to Congress on investigating internet-based crimes against children.
Please join me in welcoming Brian to the podcast today.
—--
Justin Beals: Brian, thanks so much for joining us on Secure Talk today. We're really glad to have you.
Brian Levine: It's a pleasure to be here and to talk with you, Justin.
Justin Beals: Wonderful. well, let's kick things off. You started out as a peer-to-peer networking researcher. You won a career award for it. And I'm a little curious about the realization process. you know, at what moment in your networking your computer science work did you kind of realize that some of the networks out there were a highway for child sexual abuse material?
And you needed to take that into stock in your work.
Brian Levine: Well, I really actually know the precise moment. I was at an academic workshop and I was I had organized a panel of experts to talk about peer-to-peer networking. Those in the audience who go back a couple decades or so will remember when BitTorrent came out. I think it's still moderately popular. And academics were very excited about it. The internet was still very much a novel; I should say the web was really very much a novel phenomenon, very much mirrored the excitement we have about machine learning now and the internet was changing society. And so I organized a panel on what could this new paradigm of peer-to-peer networking do for us as opposed to client-server, which really won out in the end. BitTorrent's still around, but Spotify is a major company, Netflix, all the streaming you get is not peer-to-peer. And I remember a g a friend of mine was on the panel, and he said, well, everyone knows that BitTorrent and all these peer-to-peer file-sharing networks are just or, you know, very, very much used to trade child sexual abuse material, or CSAM is the word that I'll use as we talk today. Often it's called child pornography, you know, for example, federal statutes, but it's it's more appropriate to call it CSAM since it's non-consensual. And so I was really struck by that and really stunned when my friend said that. And I felt very naive to not have a sense of that.
And then I looked into it, and long story short, I contacted some law enforcement essentially cold. I just said, I'm I'm a researcher at the University of Massachusetts Amherst. Could I speak with you? And then I got a good sense of it. And I think you asked, you know, why didn't I look away? Why didn't I go on? And it is an awful topic. It's not one that society really likes to talk about.
There's not a lot of survivors; very few brave survivors talk about this abuse that they had when they are kids. You know, we see murder all the time in the newspapers and in the movies. This is a little bit more icky. You know, it's a difficult topic. But for me, I'm not a survivor of child sexual abuse. But like many people, I've had trauma in my life. And I relate to that.
I relate to this terrible form of crime that induces some form of trauma. And I felt that there are many things that I'm bad at. I'm not good at repairing my bicycle. There are many, many things I can't do well, but I have my brain has a certain way of working with computers and computer networks, and I felt I had something to contribute. And I've been doing this ever since. And so it's been almost about 20 years that I've been contributing research and analysis, and expertise to this topic to thwart the kind of abuse that happens to children on the internet.
Justin Beals: It's always, there are these moments in our lives where our eyes get a little wider, our realization gets a little broader, and we have to adjust, you know, how we want to participate in this community, society. You know, I think that resonates. And obviously you've allowed that moment of clarity, you know, to start to drive a new passion for your work.
Brian Levine: Yes.
Justin Beals: Your lab is named Rescue Lab today. Maybe let's just start with what is rescue lab for you or or how do you how do you characterize the work you're doing? Yeah.
Brian Levine: So Rescue Lab is myself and another professor we co-direct. we have a set of professionals who build you know production quality code and help me do research. I have doctoral students who help me, master's students, undergrads occasionally; just given the topic matter, I have fewer things for undergrads to participate in. And our goal is to either provide tools that rescue kids or illuminate the harms in a rigorous way, you know, in an empirical way that happened to children online, so that we can inform people about it. Or even better, of course, we would just love it if the world was a better place, if people ahead of time who are involved in technology thought about this problem and didn't wait for it to happen. And I'm hoping that last sentence resonates with this security audience.
I think people who are professionals regularly say things like, look, you can't do security after the fact. You don't wait till you're hacked and then recover. You prepare for it ahead of time, and you attempt to mitigate, and then repair and recover. And the same is true for child sexual exploitation. You build a platform with safety by design. You think about it ahead of time. You don't react after the fact. And so we do this for a number of stakeholders. It's for investigators of this crime.
We'd like to think we're helping companies prepare for it, as I just said, with moderation practices or just best practices in general. There are a number of NGOs. There's usually one in every country that you can report CSAM to, like the Canadian Center for Child Protection, or in the US, the National Center for Missing Exploited Children. And then, of course, parents, children themselves, policymakers, journalists, and so on.
Justin Beals: Let's talk a little bit about some of the tools itself, especially along the investigatory side. I think a lot of our audience are CISOs or cybersecurity professionals. And we're always a little curious about some of the technical side of things that you're finding or opportunities to empower investigators to to find criminals. Yeah.
Brian Levine: Yeah. So I think there's sort of the the secret as you say or what what maybe what they're they're hoping that I'm gonna maybe they're imagining that I'm gonna say something like, we have this special hack, we use these zero days, and then we're able to capture people. That's not how I work. I know zero days have been used in some special operations I've certainly never been involved in. But in general, our approach is to I mean, I'm based in the US, so familiar with US law. I will rely on prosecutors and trained lawyers to really understand it. But we work with public data because in the US, you really can't do anything as law enforcement unless what you're observing is right in front of you and available to the public; essentially, that you have some authorized reason to be there. And only once you have that information, you could go to magistrate, a judge, and say, I would like something the public cannot do. I have probable cause to investigate a crime, and I would like special access into this house. And our tools are most effective when we're able to take data that's just out there for anyone to see and put a lens on it: this indicates a child might be in danger, or this indicates someone's involved in activities that endanger a child. And then they can go from there. It's not effective to build hacking tools that's just not how it works.
And to be clear, I've written papers that analyze what would happen if you were hacked, because if you can survive someone who does have special access, then your system is particularly secure. But in my child rescue work, and you know, there we focus on public data. there are cases where I would love data that are, you know, in places that are public, but I would love to get on tour forums and look at the data that's being exchanged in these forums, but the data itself is illegal. So that's another case where even though it's public, it would be illegal for me to participate in it. So, short answer is you have to be willing to look for this data and think about this crime, and then analyze it and then know about the law and know about a little, you know, and work with prosecutors, work with criminologists, work with law enforcement, and people who talk to survivors. And even so, I don't talk to survivors groups myself because there's really very few, but I work with people who do. And so you bring that all together in a multidisciplinary view. And then you're doing something rather ordinary in some sense. You're just using public data. But to do it rigorously, to do it carefully and legally and with an eye towards respect for the law, respect for people who you're accusing, respect for survivors and so on, I think you can go pretty far.
Justin Beals: How has it been working from as a researcher in, you know, higher education at a research institution? Have you found that a useful place to start to pull together that community? Because it is very multidimensional. We see this in security all the time between lawyers and technical folks and investigators. Has it been a good spot to operate from?
Brian Levine: Absolutely. I will tell you where it's not a good spot, but let me start where it's good. So a lot of the people I work with are they're doing a day job that has very short-term, pressing needs. They're working for a company where, unfortunately, their platform might be being, you know, exploited to abuse children, like leveraged in a way that exploits children. Too many exploits there. They could be investigators who are researching, or, as I should say, you know, investigating a harm against a child, and they have a caseload that they cannot ignore. And I think on TV, you know, you'll see a cop and then they go back to the police station, and there's a scientist back there who cracks the case with some, you know, either sci-fi answer or maybe even just a reasonable kind of, you know, chemical analysis of some kind. I mean, there's no law enforcement in any country that I'm aware of that has a scientist in the back room, certainly not if they're local, or, you know, even federal. Like, there are no scientists per se at the FBI. There are people who are trained that way, but they're day-to-day. Okay, so in short, people come to me, and I often ask them, I say, how can I help? And I say, I want you to tell me what you can't fix, like long-term. Sometimes they'll say, I have this technical problem in this very specific case, but what? And I'm happy to help.
But I'm more often interested in what's broken about their process, what can't they do? And then we can take the long-term view of how do you how would you address this? And that's what they don't have the luxury of doing. So, that's been a real help. And I think working for a nonprofit- I mean, a university, my university's nonprofit anyway- that there are companies that do this for profit. Forensics companies really have to meet a higher standard. They charge a lot for their tools.
I think, you know, that's reasonable to meet that higher standard. They have to pay real employees. But because I'm part of a state university, we're able to seek grants, and then we give our tools away for free. We've as I said, I've been doing this for twenty years. We've never charged for training. We've never charged for tools for the people that use them. We seek grants from the federal government, or we seek funding from foundations, or we seek gifts basically. And I think doing this all together from my vantage point is great. And, you know, to have the time to teach and really think in the long term. I think where it's not great is it's just, you know, maybe there aren't too many bad things, but I think it's hard, it's hard to keep this together with funding, I guess I would say, is the real challenge.
Like I spend a lot of time trying to find funding for what we do, and I don't mind explaining what we do to receive the funding, but it can take a lot of time. And maybe if I was at a company and we had a business model where we could do this for a profit, that would be smoother. But I expect the grass is greener, and they have marketing departments and all sorts of things that sales are very important and so on. So all in all, I think it's a great perch to be on, a great position to be, maybe I shouldn't say perch because I view this as a multidisciplinary, all parties are equal kind of effort.
Justin Beals: Yeah, it's an interesting philosophical tension. I mean, and there are layers to it. You, you know, in your in discussing how you became inspired to do this work, you talked a little bit about these peer-to-peer networks that we were building. I was around at that time. It felt like we were breaking down distribution systems that had been put in place. And for someone that didn't have access to that distribution system, being able to distribute material, content, information without it.
Brian Levine: Yeah. That's right.
Justin Beals: Was supposed to be revolutionary and things that people could learn and have access to. At the same time, you know, there's this commercial tension, you know, of productizing and centralizing that data. And we see that in social networks these days, right? Compared to something like Mastodon. Yeah. please, yeah.
Brian Levine: Yeah, could I speak to that for a minute? Is that okay? Yeah, I think there's a couple different tensions. I think what happened with- I remember, and I agree there was this view of disruption of old paradigms with peer-to-peer. And I think what it failed to do in, you know, everywhere peer-to-peer has shown up, is you can't provide customer service. That you just you, you know, you just that's an important part of business models that can't be provided.
And that's just one side of it. The other is you can't moderate. There's just like you can't fix problems, repair them and monitor the network very well, and there's no one in charge often. It's also the case that you can't control your users and be a responsible user. And I think the old-style 1970s- I think it's Steve Levy wrote a book called The Hacker Ethic, where information must be free is really outdated and irresponsible, and I don't think that's the way we should be running society in these days. And, as you said, I started my career doing peer-to-peer, and I quickly realized what a mistake. What a poor way of doing things. So.
Justin Beals: Yeah. I have to tell people, yourself included, that I worked in education for a long time, and I don't anymore. I had similar type of moments where I was like, wait, what are we building this software for? I don't think it's achieving what we wanted out of it. And it did leave a sour taste in my mouth because I certainly we all need to make a living. We all want to provide for our families, and at the same time I don't want to make the world a worse place. I'd rather improve it in some way. Speaking of like peer-to-peer networking and especially Tor, wanted to dig in there because you definitely delineate a little bit on Tor as a system for good or used for bad. These are tools. And I think you have a delineation around the Tor Browser versus Tor Onion Services and some concept of how they're being used and how we should perceive them.
Brian Levine:
Exactly. Yeah, I do. And I think it's important to make that distinction. As you said, there's Tor Browser, which is, you know, probably a large majority of your audience already knows that it's a different kind of VPN where you can connect, I'll call it, to the clear net, the regular internet. You can go to any newspaper, any government site, any company. And what its goal is, is to make it hard for the site you're visiting to know your real IP address. And that's it. You know, if you tell that site this is my real name, well, that would be stupidity, right? It doesn't make sense to, you know, meet someone with a mask on, hide your face, and then tell them your name. And so that's the same with Tor. Now the other product they have is called Tor Onion Services. And that's when, for example, you're trying to provide a website and not reveal your IP address.
And so similarly, it wouldn't make sense to be a large company to provide a Tor Onion service, hide your IP, and say This is the name of my company. So they have very different use cases. The reason they're joined at the hip is they rely on the same set of volunteer relays that different people or organizations around the world have put up. And I think their use cases are very different.
When you contact a Tor Onion site, you are using the Tor browser. So your IP address is hidden, and their IP address is hidden. And then you have to start to ask yourself, what's the use case here? And we could dig into that. But I think that the use cases have different ethical quandaries for sure.
Justin Beals: Yes, 'cause it's a little tough, I think, because even when we had a conversation with this with a gentleman named Ben Collier, about the invention of the Tor browser and onion services, and it's an interesting dichotomy because on one side he's saying, Well, we built this for anonymity, privacy.
But at the same time, we realized that the CIA was funding us to build it because they wanted to be anonymous in their work online. And their motivations weren't always aligned with altruistic behavior. You know, they might have been more motivated by a nation-state opportunity or a directive that they were receiving. And they had to be willing to allow both to get the level of anonymity or even the funding to build the solution.
Brian Levine: Yeah, I think that's right. I would argue they're not providing high-quality obfuscation of anyone's IP address. I don't think it's a secure product. I think it's, you know, I don't know the CIA's motivation. I think it was actually the Navy research lab that Tor came out of. Yeah. And MIT and some others. And it's been worked on by academics across the world. It's a nice It's a problem that models well with math. And I think that's been appealing to people across the wheel across the world. And I think it, you know, on the face of it, has this great purpose of, let's provide people the ability to be private online. I think that's important for adults to have that. But I think in the end, there's a number of problems that they control a network, that they, you know, control who's part of it. They've kicked people out; they control the code. It's an open-source product, but you can't submit code without it being approved. And then they've introduced features that protect people on the network who volunteer for them, but they have no features for others who are exploited because they've released Tor. And I think there's an opportunity for them to improve the network to prevent that exploitation of others, but they seem to refuse to want to do that.
And I think it's an extremist position to say all speech, even speech that hurts others, particularly malware, ransomware, and child sexual exploitation abuse material (CSAM, and to allow that and then not even to provide privacy very well at all. I mean, honestly, I think it's a piece of junk. And I don't think people say that often enough with Tor. They don't say that it really does not provide the level of security that the movies or the website purport to provide. I just think all of it is a waste of time, and a lot of people are getting hurt because of it. In and in particular just for just for the onion services, I wanna be clear here. I'm not talking about Tor Browser. Although Tor Browser is not that great either in terms of protecting your privacy. I just don't think it hurts as many people, for sure.
Justin Beals: I have to imagine that things like that, we've seen this a number of times, kind of being in the wild are not necessarily in the wild without having been allowed to be in the wild by someone that can reverse engineer the system. I think in this internet age, you know, the transition here has been like, I don't think we were as worried about some of the impact information or communication or speech has on people prior to so much speech being online. It has been a ground shift, it feels like, culturally.
Brian Levine: I think, yeah, I think that's true. I think people predicted the world would change from the ubiquitous access to the internet and information. I think, you know anyone short of visionaries couldn't have imagined where we are now, particularly with these AI models ingesting so much information and processing it so quickly. But I think, despite the fact that the world has changed in unpredictable ways, the limitations of the Tor project are very clear, and the fact that they're not taking action to address them is questionable to me. And I don't think there's much; I don't see much lack of clarity as to any kind of grey area that they're in. I just don't think they provide a good product, and I think people are getting hurt, and they're essentially a public nuisance in my mind.
Justin Beals:
Yeah, and people are getting hurt. maybe I'm I'm off the beaten path here a little bit from some, but I'd love to ask can you characterize the scale of the problem as you see it? CSAM specifically, in the work that you're doing? Yeah.
Brian Levine: Yeah. Yeah. So they're they're really, you know, exploitation of children has been around since before the internet. And, you know, nothing has changed its character more than the internet and the platforms that are available. There's not, you know, a type of communication or type of platform on the internet that hasn't been leveraged, you know, to exploit children. People may remember newsgroups or Usenet that used to be pretty heavily involved in that.
From early on, the web has been email, search engines, I mean, you name it. Certainly, social networks now are and Tor and all the peer-to-peer networks that are there. And the thing is, you know, it's an endemic problem, and we shouldn't, we should be doing two things. We should not be triaging the problem. The word triage, to me, says, we had this unplanned for spike of events that happened, and we've got to quickly put resources together until things subside.
This is not going anywhere. And so, since that's the case, as I said before, when you have a platform, be it for-profit or nonprofit or open source, you should be designing it, expecting people to abuse it to exploit others. And in the case of the Tor project, it's the basis of ransomware today, right? It doesn't cause ransomware. It doesn't create ransomware, but when you are subject to ransomware, that ransomware is using Tor as a command-and-control platform, right? It's being used, you know, it's part of ransomware notes everywhere. It there's no obvious good side of ransomware that I can speak to, at least in an ethical fashion. It can make some people more wealthy. And then similarly, and I just you know, it should be filtered from the Tor network. That's entirely possible. But they won't do it.
And then similarly, I don't know of an ethical use of CSAM. It's, you know, it's again of no use to anyone. Now there are, and this is in every society, on every platform, and definitely rampantly use both of these on the Tor project software, particularly onion stuff. And they're aware of this. They, you know, the Canadian Center for Child Protection has a website where they talk about how they send certified mail noting the problem to them. It's every security researcher should know that Tor onions are used as an essential part of malware, of ransomware. And I just think it's such a part of our society that it's questionable to me that nothing's being done about it. I think the other thing I'd say is just to say I don't know why they don't reveal that. I don't think they're transparent about these harms. And I think I also think that we should, well, I guess I'll just leave it at that.
Justin Beals: Well, I think there's a cultural dichotomy in the computer science world around this, right? Where I think their concept that they're just the infrastructure, they're they're not running it. We see hear the same thing on the corporate side. It's not different from a nonprofit like the Tor group versus a social media network that is like, hey, we're hey we're just the medium, not the media. But they control the medium, I think, is what you're pointing out. We think that they don't, but they they actually do by making the decisions that they do today.
Brian Levine: That's right. And I think it would be a good exercise for anyone who doubts this to compare, say, the Tor project to BitTorrent. There is no BitTorrent project. You can create your own version of BitTorrent right now. You can modify it in any way you want. This is for people who don't know a file-sharing network. And people have done that. There have been modifications to BitTorrent. And when you release your version of BitTorrent, there's no one to kick you out because nobody controls BitTorrent. If you modify Tor in such a way that you do not pass the daily tests that they run against nodes who join. You get kicked out, and they have kicked people out. So they control what is considered good behavior. If you were to filter in a way that they don't approve, they would kick you out. So it is again open source, and it's run by people over the world, but it's managed by a small set of people on the Tor project and a set of five people who are really active in a group called the Directory Authorities.
And like I said, they're they're aware of these problems. They appear to me to be refusing to fix it. And it's been going on for decades. And then again, it's not even that good in terms of protecting the people they purport to detect protect, whistleblowers, journalists, and dissidents. We haven't really gotten to that yet. And I just think that's a public nuisance, right?
I'm not even complaining about the content that's there. I'm complaining about their operation. If you have a, you know, if you have a factory that is belching smoke everywhere and they know the problem can be fixed and they've been told that it's a problem and there are fixes, but they say it's all right to, you know, expel this much exhaust into the environment, then I would call that a public nuisance.
Justin Beals: Yeah. Yeah. Let's talk about some of the lexicon of words that we like here. Public nuisance, that's a good one. I think for as much money as some companies make, I might perceive them as a public nuisance as opposed to a benefit. But also, I think you like to point out that words like dark markets or privacy infrastructure kind of create a shield around some of these systems so they protect themselves from having to respond to what is wrong that happens on the things they control.
Brian Levine: Yeah. I particularly don't like the term dark web or dark market. And the reason is, you know, it doesn't even really mean anything. I think the origin of it, one origin of it, is that it's content that Google doesn't index. I imagine, you know, the AI models similarly are not ingesting much content from the onion sites that are on Tor.
And there's really, There's only, you know, one to three different anonymous networks out there or dark web that are out there. And Tor is the 800-pound gorilla that's out there. It's really all of the traffic. And I think when journalists say dark web, they're doing it because people have some notion of some vague notion of what that means. And I think they should just start saying Tor markets. Or they should say, you know, this website that was on Tor. Because these are websites that are only accessible on Tor.
And the only reason that Google does not index them is because they clearly could. I mean, why can't Google index them? Is they disagree with the content because it's often content that is, you know, harms against children, let's say. And there's no need to index a ransomware site. That's just to the benefit of the ransomware site. So I think, yeah, I think you're right that we should choose our words carefully, and we should ascribe responsibility, if not responsibility, at least be precise about the fact that it happened on tour. And I think they allow it because that has entered into our lexicon; they take credit for things, and then they don't take responsibility for things that also happen on the same technology.
Justin Beals: Yeah. I like your point here that these are indexable because I think that the dark web concept makes it feel like, well, it's mathematically impossible to understand how much of is out there. But that's not true. These are still fundamental IP addresses on servers, you know, hosting a different type of service than HTTP, that could be crawled just the same way they run through all the domains, you know. It's not impossible for them to investigate it or surface it if it was useful.
Brian Levine: Correct. And there are search engines you can find on Tor onion sites that index these websites on Tor. They're run by single people. You know, Google's one of the richest companies in the world. And to think that they don't have the ability to index these sites; they're doing it by choice. it's in fact, interesting to me that they made the choice not to do it, and they do index other sites that are harmful to society. But anyway, that's a decision they made long ago, and they've stuck with it.
Justin Beals: Yeah. This internet we weave a little. Right. Yeah. one thing that I wanted to dig into a little bit, you mentioned dissidents, you know, it's often used as a way to describe I think back to the great firewall where Chinese dissidents might want to be able to discuss more openly a as a use case for a system like Tor. How do you think about that type of need for anonymity and privacy, a whistleblower type thing, on the internet today, or if not Tor, what would be your recommendations around work?
Brian Levine: Right. Okay, so that's a great question. So I'm all for protecting, you know, the protecting those who are vulnerable in our society. That can be, as you said, whistleblowers, dissidents, could be reporters, and, you know, who are who are exposing certain bad actors. And I think they can use Tor Browser to connect to the regular internet. There's nothing that's happening that you couldn't do. There's no website you could contact on Tor Onion sites that you couldn't contact to help these people that you couldn't contact with Tor Browser. So, for example, people often talk about SecureDrop, which is a way to contact reporters at different prominent newspapers to offer tips. Like I said before, if you're offering a Tor website, a Tor Onion site, and you say, I'm the New York Times, it doesn't have to be an onion website. Just put it up on the regular web and say, if you have a tip, contact us with Tor Browser.
So we don't need to; that's why I separate out these onion sites, which protect also people producing malware, ransomware, and people who are trading child sexual abuse material. there's many different things in the world, and some of them are illegal in some countries and legal in others. But at a base level, there's not anywhere where ransomware is legal, or the extortion of money from people is legal, and there's not anywhere where child sexual exploitation is legal.
And we're not talking about, you know, kids who are 17, I mean, child sexual abuse goes down to infants. these are clear-cut cases of just terrible, awful abuse. so I th I think it's important to protect these people. I think you can do it in t with Tor Browser, and you don't need to keep this other product that is just helping much worse people. The second thing I'd say is I wouldn't use Tor Browser for this. It's really not that good.
You know, one of the deaf when I teach security, I tell my students that one way to define security is that what you are trying to protect, what when an at when an attacker would like to gain access to it, they essentially have to spend more money than it's worth to them to gain access. If you put a thousand dollars in a safe and it costs them a thousand and one dollars to break into the safe with, like, a really good drill, then a rational attacker would not proceed. And so if you think about who is after dissidents, and whistleblowers and reporters, it's governments. I mean, there is not a government in the world that cannot defeat Tor. And I think one of the problems with the Tor project is they have not quantified for everyone just how much money you need to break Tor Onion Routing or even Tor Browser. But let's just stay with Tor Onion sites. And I have done that. You can go to my site, which is Tor-harms.github.io., and you can see my simple model of how much money it costs to break Tor. And we're talking like tens of thousands of dollars. And so I often quip in my class that Tor onion sites are good for protecting you against maybe your local delicatessen who doesn't have money to spend on Tor. But for major corporations and governments, this is really ridiculous to risk these people's lives with a product that doesn't work. And I think the last thing I'll say is just it does use encryption. And I think people make this assumption that, well, it's got cryptography involved. And I know it's hard to decrypt things that are properly encrypted. That's not what's going on here. You don't break the encryption of Tor. You disrupt its routing through the network. And it's a very weak system. There's no revelation. I'm this is not breaking news. These are just things that are available in papers since the early 2000s, when Tor came out, and I just think it's a shame that they do not transparently define, they do not ask their users, before you turn on Tor, ask yourself, does your attacker have this precise amount of resources? And if they do, we can't protect you.
Justin Beals: Transparency, right?
Brian Levine: Transparency. Right.
Justin Beals: So yeah. And I think what's absolutely intriguing to me is what would be the technology hype engine. And I feel like we're in one of those cycles again. And it's like this technology has been hyped to high heaven about its privacy. It's, you know, and sometimes poorly. I think about the Silk Road. There's been some high-profile instances where, yeah, dark websites have been found and shut down. And I it's funny how we think immediately because we've been marketed too well that well this will make me anonymous. But in truth, you should probably second-guess it 'cause it's a target for those that don't want anonymity.
Brian Levine: That's right. I mean, it's, you know, the the question you should ask yourself when you want to secure something is what is the level of attacker required to get around the security? There's always someone. And to offer a product that doesn't quantify that is bad; it's just bad in the commercial marketplace. And it's no different for an open source project as well.
Justin Beals: Yeah. You have helped express to us your passion in helping reduce CSAM or harm to minors, children in your work. What does the new work look like? And I'll give you one thing that's just risen to my brain. I'm rewatching The Wire lately, that old TV show about Baltimore. And every season has a new network.
Brian Levine: I love The Wire. Love The Wire. Yeah.
Justin Beals:
That they have to figure out how to listen into. And I have to imagine your day in, day out must feel kind of similar. Yeah.
Brian Levine: Right. Yeah, I think it's true for when I started doing work in thwarting child exploitation. It used to be that operating systems didn't come with encrypted file systems. It used to be Microsoft had something, but it was hard to enable. A lot of people used TrueCrypt, and there were companies that would sell drives that were encrypted. and then one day all the operating systems said, we're just gonna enable encryption by default.
And some people are like, my God, this is the end. and so they're they're, you know, we we moved on, and it is true; sometimes they get to a house where they can't decrypt things. but you know, but you persevere and you continue. And I and I think machine learning has definitely changed things. Deepfakes are a real problem these days. for example, just using Nudify.is a problem for adults and children. It's a tactic of abuse and coercion and harm and exploit you know extortion. And then also, even generating synthetic CSAM content is a harm because it can lead law enforcement on, you know, a rabbit hole, a goose chase for someone who doesn't exist, and it also normalizes the content; if you can see that much more of it, it's a real problem. So I I think I think AI is, you know, probably to no surprise to anyone listening, the the way it's been rolled out without any real guidelines from the government, often with not enough guardrails. There there are news stories everyone can look up about how chatbots talk to kids in the wrong way, or you know, talk to adults in the wrong way that normalizes things and and deep fakes and all that, that's really the latest concern, et cetera.
Justin Beals: Yeah. Always a new technological frontier. someday maybe we'll learn the lessons secure by design. I like that you brought it up in the beginning. Yeah. I think we can all agree, those of us that are operating in this security space, that thinking about the harm that can be done, it there's not a lot new under the sun there. You know, that this has been going on for a while. We could consider those issues in our design phase. Yeah.
Brian Levine: That's right. And I and and I think long ago there used to be this rag to riches story for, you know, especially programmers and computer scientists. Like, wake up from a dream and you say, my God, I've got this app. I'm gonna put it on the App Store. Tomorrow I'm gonna start making money. Later in the week I'll be rich, and I'll be bought out sometime later. But we're at the level of maturity as a society right now and as a profession, that part of your plan should be not just the coding of the app and putting it on the App Store, but some plan for moderation of the content, reporting of the content, thinking ahead who can exploit third parties that are not even part of your platform. It's not, it's it's very outdated these days to say, I've got two people talking through my app and their communication is secure. Well, who are they hurting by having that ability to talk? That's that's part of your design phase, and it should be any responsible company should be doing that at the startup phase.
And there's no time; it's never too late to start if you're not, you know. If for those listening.
Justin Beals: Yeah. Yeah, absolutely. I I I can also tell you as someone that has built a lot of tech startups that the odds of having that epic outcome are way against you. it it is a real challenge, exciting challenge to both develop a great technology and a great business and also bring it to market. But it's work I've enjoyed doing. hopefully I get better at building things that make people's lives better all the way around. Brian. Yeah. Brian, I am I I yes. we need more of that. A little more ethics in computer science and business, I think would be helpful.
Brian Levine: Absolutely. Yeah. That's everyone's goal, I think.
Justin Beals: I'm deeply grateful for your work and am appreciative of your passion and kind of the foundation from a researcher that you start with in in higher education. Brian, thank you for joining us on Secure Talk today.
Brian Levine:
It’'s been my real pleasure to talk with you and have the chance to connect with your audience and I hope to talk to you again soon one day.
About our guest
Brian Levine is a Distinguished Professor in the Manning College of Information and Computer Sciences at the University of Massachusetts Amherst, where he is also Associate Dean of Research & Engagement and Director of the UMass Cybersecurity Institute. He received a PhD in Computer Engineering from the University of California, Santa Cruz in 1999. In 2020, he was elected as Fellow of the ACM, “For contributions to network forensics, security, and privacy, and for thwarting crimes against children.” In 2022, he was the sole author of a Report to Congress written on behalf of the National Institute of Justice on the investigations of internet-based crimes against children. His work has been featured in the New York Times, Wall Street Journal, and other major venues.
Justin Beals is a serial entrepreneur with expertise in AI, cybersecurity, and governance who is passionate about making arcane cybersecurity standards plain and simple to achieve. He founded Strike Graph in 2020 to eliminate confusion surrounding cybersecurity audit and certification processes by offering an innovative, right-sized solution at a fraction of the time and cost of traditional methods.
Now, as Strike Graph CEO, Justin drives strategic innovation within the company. Based in Seattle, he previously served as the CTO of NextStep and Koru, which won the 2018 Most Impactful Startup award from Wharton People Analytics.
Justin is a board member for the Ada Developers Academy, VALID8 Financial, and Edify Software Consulting. He is the creator of the patented Training, Tracking & Placement System and the author of “Aligning curriculum and evidencing learning effectiveness using semantic mapping of learning assets,” which was published in the International Journal of Emerging Technologies in Learning (iJet). Justin earned a BA from Fort Lewis College.
Other recent episodes
Keep up to date with Strike Graph.
The security landscape is ever changing. Sign up for our newsletter to make sure you stay abreast of the latest regulations and requirements.

