- Welcome to another episode of "Supply Chain Frontiers," the MIT CTL podcast, where we explore the trends, technologies, and innovations shaping the future of supply chain management. I'm your host, Mackenzie Berry. Today, we're discussing cybersecurity and how it impacts supply chains. Joining me are researchers from the MIT Center for Transportation and Logistics, co-founders of its new Supply Chain Cybersecurity Lab, Doctors Miguel Rodriguez Garcia and Camilo Mora. Thank you both for joining me here. First question for you all, what is the Supply Chain Cybersecurity Lab, and why launch it right now? - Let me tell you a quick story, Mackenzie. So, in my case, about like a year and a half ago, I was at an industry conference when a colleague of mine, a supply chain professional, came up to me and said, "Camilo, we were attacked." Of course, my first reaction was like, "What do you mean, like, you were attacked?" And then, this person explained that their company had experienced a ransomware attack. The incident disrupted operations nationwide, and it took them, you know, roughly, about, like three months to fully restore and recover their systems and operations. That moment, that conversation really stayed with me because it made me realize, like the enormous impact that a cyberattack can have beyond the classic IT department. When we think about supply chains, we normally talk about three fundamental flows: the flow of goods, the flow of information, and the flow of money, the financial flow. A cyberattack can certainly disrupt all three at the same time. And on top of that, you have the potential reputational damage and impacts on customers, on suppliers, employees, you know, your entire network. Shortly after, Miguel and I started interviewing companies that had actually experienced cyber incidents, and something became very clear: the classic problem, organizations working in silos. So, on the one side, you had like, you know, the supply chain people. On the other side, we had the cybersecurity and IT teams. Both dealing with, you know, the same disruption, but from completely different perspectives. That is when, like, Miguel and I saw the need to create the Supply Chain Cybersecurity Lab. Our goal is to understand how cyberattacks translate into physical operational disruption and then help organizations become better prepared to prevent, mitigate, and respond to those disruptions. - For listeners unfamiliar with supply chain cybersecurity, why should non-IT leaders care about this issue? What makes cyberattacks relevant across supply chains today? - I started working actually on cyber and kind of like the connection with supply chains about three years ago because we had a pretty big project with the U.S. government to basically find different vulnerabilities in the warehousing space when it came to technology risks. Of course, cyber was the biggest one that came out of that study, and because of that, we realized, "Wow, cybercrime, mainly after COVID, was basically going through the roof." Just in 2025, to give you an idea, the impact of cybercrime in the world was estimated between $2 trillion and $10 trillion, with a T, right? Just the average of that would be $6 trillion. That's the third-largest country by GDP, only after the U.S. and China. So, the scale, you know, of cybercrime worldwide was huge, and the targets were changing. I think most of the people in the audience will think about cybercrime and maybe, you know, they are gonna envision a guy in a room, you know, like a teenage, like, kid, like back in the, you know, 90s or so when the internet was started. That's not the reality anymore, right? - Yeah. - We have a bunch of organized crime. We have a lot of state-sponsored, like, cybercrime nowadays. Also geopolitics is not helping in that sense, unfortunately. But the reality is that because of that, we have seen a change in how the attackers are actually performing the attacks. So we see more targets that are just aiming at disrupting operations, and that means a lot of the times stopping physical supply chains, like disrupting a manufacturing plant from producing, a warehouse from sending product, and so on. So that is why nowadays anybody in the operations world and in supply chain in particular needs to think about this in a different way, right? Because these attacks are already impacting us, and it's huge. - So there's a very high level of sophistication, - Yeah. - That we're looking at. - So super high level of sophistication when it comes to the new attacks and, at the same time, unfortunately, a very low level of sophistication a lot of the times on the operational technology layers of our supply chains, and that's 'cause we rely still a lot in legacy systems. A lot of them, if you think about warehouses, manufacturing plants, even vessels or ports, we've had pretty, like, I would say, good automation systems, like physical automation systems, for like 50 years or more. Like after the Industrial Revolution and after World War II, for instance, manufacturing was the first sector that moved quite fast in that sense, right, in terms of automating a lot of the process. But those robots were not, you know, supposed to be on the internet. Nowadays, they are, but you cannot patch them a lot of times. So a lot of the traditional operations are now super vulnerable to modern attacks. - Yeah, I agree with that. So, like, essentially, people who deeply understand cybersecurity do not necessarily understand how a warehouse operates, how transportation networks work, what happens on a production line, you know, or what, like, the operational consequences are when a critical system goes down. And on the other side, supply chain professionals are not always trained to understand cyber risk. So we need people who can speak both languages, right? We are, I would say, trying to help companies continue operating even when technology fails. - And it may seem obvious, but in terms of the motivation that attackers have for doing cyberattacks, primarily money, primarily wreaking havoc, a mix of both? - So we have seen a change in that. Like, for many years, money was the main one. Ransom is still the main type of cyberattack that basically encrypts all of your data and locks your systems, in a way. And even some of the recent ones, they're still ransom, but they are not actually asking for any money back. Instead, they are just trying to disrupt the operations anyway. And that has to do with the fact that we are seeing more state-sponsored cybercrime in the last couple of years. So that is unfortunately also a big change in, yeah, in the motivation of malicious actors. - I wanna speak to a specific example that you can walk us through. Can you walk us through a recent cyberattack and its operational impact, just to give us a frame of reference? - Yeah, sure. There are many, unfortunately, but I think 2025, the Jaguar Land Rover attack was one of the, I would say, biggest ones in history, but also more, let's say, educational, right? This was an attack that basically targeted the car manufacturer Jaguar Land Rover in the U.K. They had to shut down operations, like, in all manufacturing plants in the country for five to six weeks. But the thing was that not only was, you know, Jaguar Land Rover affected, but when you have to shut down, you know, for six weeks production, like, your tier one, tier two, tier three, tier four suppliers are impacted very, very rapidly. And the reason for that is 'cause also a lot of those suppliers, usually they are, you know, small and medium-sized companies that have basically no, you know, cash runway to keep operating without getting paid for a long period of time, right? So the total impact actually for the U.K. economy of that single attack was between $2 to $3 billion on a single attack to a single company, right? - Billion with a B. - Yeah, billion with a B. For Jaguar and Land Rover specifically, just the cost of cyber recovery, which is basically when you have to, you know, call consultants, lawyers, and usually, like, a recovery battalion, cybersecurity team that just, you know, goes in when the worst happens. They spent more than $200 million just on that, and that's not accounting for all the revenue losses that were estimated between, like, $50 to $100 million per week of lost production. So it was huge. They had to reset everything, basically isolate all the network and basically start almost from scratch. But yeah, it took them months. Until mid-November, they were not back up. I wanna mention now 'cause one of the reasons why we launched the lab is 'cause actually there is not a lot of visibility when it comes to all of these attacks impacting the supply chain. And one of the first things we did actually was, "Okay, how do we bring this to life?" Right? So we created a tool, and it's called TRACE. Camilo, did you wanna maybe explain a little bit about what TRACE does? - Absolutely. So it is very important to learn from those attacks, you know, like, the case of Jaguar Land Rover or Asahi or many others. But usually, what about the lessons that other companies can learn from those incidents? And not, like, isolating them, but you know, like, to look at them globally. So as Miguel mentioned, there was no, at the time, or to the best of knowledge, a platform with structured data about the incidents that affect the supply chains. And then, you know, we decided to develop TRACE, which is the World Supply Chain Cybersecurity Observatory. It's free-to-access platform where we kind of collect and organize publicly available information about cyber incidents that, in particular, have affected supply chains around the world. And the idea behind it is very simple. We wanna learn from the past so that we can better be prepared for the future. For each incident, we try to capture and understand not only what kind of cyberattack it was, but also what happened to the operation. So which industry was affected, which part of the organization was hit, was manufacturing interrupted, how long did the disruption last, what systems were involved, so on and so forth. And if the information is available, also, what was the financial impact of it? - One of the issues that we have is that a lot of these attacks are not publicly available, - Mm-hmm. - Or publicly disclosed, right? - That's what I was gonna ask you about 'cause no company, of course, wants the PR of, - Yeah, yeah. - You know, "We have a cyberattack." - That's still a limitation. And that's one of the reasons also why we wanted to create the new initiative, right? 'Cause we wanna learn from a lot of these companies that have suffered, you know, these cyberattacks. And I think it's funny that you just said, for instance, like, you know, the PR side. I would say maybe 10 or 15 years ago, that was a good reason for it. Nowadays, everybody, you know, gets hacked. - Yeah. - So yeah, you might have a, you know, short loss in, you know, your stock price for a week or two, but also there is a lot of research now that shows that that usually recovers pretty fast, 'cause as long as it's not one of these massive, very harmful attacks, the impact on the operations might not be, you know, that high. And those, we know them, 'cause those, they are disclosed- - Sure. - No matter what, right? So, but yeah, there's still a lot of maybe attacks that have, you know, like, a medium to small impact that probably we're not aware of. - Yeah, and you know, there's a lot changing, as you all mentioned, in cyberattacks. It almost feels like you have to be ahead of the curve or at least on the pulse, because if you're behind, then you're vulnerable. Can you all speak to, you know, what's changing in cyberattacks on supply chains right now? - I think one of the biggest things changing is the attack surface. So supply chains today are much more connected than they were even a few years ago. Companies rely on cloud platforms, software providers, technology vendors, connected equipment, suppliers, logistics partners, so on and so forth, and of course, AI. Every new connection creates value, but that can also create another potential point of vulnerability. So AI, in fact, is a very good example. Right now there is almost a race amongst companies to implement AI. There is this perception that if we are not doing it, our competitors probably are, right? And they're getting ahead of us, so you know, they're scared, and they want to implement and, you know, launch and develop. So some companies are encouraging employees to experiment with AI and become more productive, and that is good, you know. That many, like, innovations may come out of it. But the challenge is that the adoption can sometimes move faster than governance. So a company might officially implement, I don't know, Microsoft Copilot, for example, while employees are also using their personal ChatGPT, Claude, and other AI accounts. They may be using them for personal devices, uploading documents, analyzing data, you know, connecting, you know, different applications, or even developing things. And suddenly the organizations may not even have complete visibility into which AI tools are being used, what information is being shared, or where that information is going, you know? And again, AI is just one part of it. The other one, OT. - Yeah, I mean, OT, as I mentioned before, the operation technology side is a big part of it 'cause we have seen a huge proliferation of technology providers in the space. But the reality is not only on the OT side, it's also on the maybe IT side of supply chains, right? If we think about, like, maybe 10, 15 years ago, and e-commerce had a lot to do with it, the amount of technology suppliers was pretty much, like, I would say, a small number. The big suppliers of supply chain technology were a few. You could list them across in a warehouse management systems, transportation management systems, of course, you know, the big ERPs and so on. Now there are so many providers of everything. Even, you know, if you think about robotics technologies that, like, in warehousing, which is my background, like, again, 10, 15 years ago we had just a few that were kind of reliable and were, I would say, you know, top tier in the field. Nowadays you have hundreds of them. So that proliferation of, you know, suppliers, usually small, like a lot of them are even startups, right? That means that most of those companies do not have the cybersecurity, you know, capabilities that a lot of the big players have. That is a huge, you know, open door for cybercrime. And at the same time, we're implementing, again, more technologies. That means, as Camilo mentioned, that we have basically a bigger attack surface, right? As you have more technologies interconnected, your attack surface just increases. Attack surface meaning where the attacks can come from, right? I remember now that one of the main logistics providers worldwide, having more than 250 different technology providers for the warehouses worldwide, even though each of the warehouses usually uses only between 10 to 20 different technologies. But still, because of course every facility is different, you end up with 250 different. Are you gonna be able to really control all of them to, you know, audit them from a cybersecurity perspective? Of course not. It's super hard. And also for attackers, they know that, right? And these tech providers are the weakest links. So that is one of the, I would say, changing, like, fastest, you know, changing things that we're seeing now in supply chains with modern technology, that technology providers are becoming the new target for cybercrime 'cause once an attacker hits one of them, basically the impact then cascades across the different organizations too. And there are many examples of those. I would say maybe one that I suffered directly last year was a pretty good example 'cause it happened in Europe, and it happened to many different airports at once. - Yes. - 'Cause the idea of how can you actually shut down more than one airport at once, well, I mean, probably we couldn't do it, like, 10 years ago. But now these guys, they basically target a small supplier that nobody was paying attention to probably, that was handling the check-in and boarding processes. So a small software supplier, and basically they had to freeze all the check-in kiosks. They halted all the, you know, automated baggage tag printers across more than four airports in Europe. I think it was Brussels, Berlin, Heathrow in London, and I think Dublin in Ireland, so. - And was that one just for havoc? There was no ransom. - So that one was also ransom, but again- - It was. - They didn't ask for, like, a payment again. So yeah, the goal was disrupting operations, yeah. - And in manufacturing, something similar is happening. So many companies still operate equipment and legacy systems that were designed 10, 20, or even 30 years ago. And originally most of those systems were relatively isolated. They were not designed to be connected to the internet, to the cloud. So today those systems are increasingly connected to IT systems, cloud environments, vendors, as you were describing, and remote access tools. So you have old technology operating in a new, highly connected environment. And as you said, Miguel, attackers are using that. - Yeah, yeah, yeah, and you cannot patch them, most of the time, so. - Our vision, just to give, like, an overview of our research approach and how we are translating that into certain, like, tools, is, like, we have our research agenda. We wanna learn from the past, so for that we built TRACE. You know, look at the past, learn from those incidents, et cetera. We are building another one, which is RADAR, to anticipate the future. Essentially it's to identify emerging risks, threats that could affect supply chains from the standpoint of cybersecurity. And by connecting the inputs or the outcomes from those two, we are building SCOUT to shape the present. So SCOUT stands for Supply Chain Operation and Vulnerability Threat Assessment. And companies have, you know, hundreds or even thousands of unknown vulnerabilities across the organization, and also upstream. So the idea with SCOUT, imagine a platform where your employees actually interact with it, learn by interacting, which is another key piece, and also you can get a quick assessment of your company across your different functional areas and levels. And with that then, like, determine a roadmap, prioritize what to do next, you know, and also being, like, dynamic because again, like, technology is nowadays changing so fast that the traditional classic assessment, you know, checklist that they do once a year, you know, it's outdated. Even if they finish today, maybe tomorrow it's already outdated, you know? So that is a SCOUT. It's a tool that we're creating for organizations. Maybe large companies have their own vulnerability management platforms or systems, but what about the tier one, two, three suppliers? Maybe they have some certain visibility and control on their tier one suppliers, but, like, in terms of compliance, but what about the others? - You know, we've talked about how it needs to be beyond IT teams within organizations looking at cybersecurity. What kinds of roles or teams are you all envisioning as using something like SCOUT? Are we looking at operations teams? - Everyone. Yeah. Everyone across the organization. In terms of cyber, in terms of security, it's a matter of everyone. As Miguel was mentioning, you have the case of the airports, you know, and maybe, you know, you kind of create certain barriers or kind of protect key things such as, you know, the radar or, you know, like navigation systems of the planes, whatever. They never thought that the way to create a disruption was, you know, that you cannot check in luggage, right? - As simple as that, yeah. - As simple as that. - So we've talked about some things that are changing quickly in cybersecurity and some of the tools that you all are offering through the lab. Can you give us a kind of overview of the current challenges that companies face in supply chain cybersecurity right now? - We look at, for instance now, supply chain cybersecurity in two kind of like axes, and that's something that we started doing 'cause I like the theoretical part of research, right? So I like to read from other different fields and bring ideas together. So when I was thinking about cyber and the scale of these disruptions, I went back and read about, for instance, like, you know, nuclear reactors basically having issues and creating some of the, you know, largest disasters that we've had on Earth, but also spaceships going burst, which is also something that is very, very hard to, you know, to control and to predict, right? And when you launch something to space, there is so much uncertainty in what can go wrong, right? So what we realized is that when you look at supply chain cybersecurity nowadays, it's moving across two axes that are related to that. One is the level of what we call coupling, and coupling meaning the interdependence of systems, right? When you think about, again, a spaceship that basically goes, you know, to space, like when something goes wrong in one part of that ship, it's so easy for that issue to basically then, you know, travel to any other part of the ship. Like if something starts heating up, the next, you know, metal part is also gonna be heating up, things like that. Nowadays in supply chains and cybersecurity specifically because of the amount of interconnectivity that we have in IT systems, but also those IT systems connected to the, you know, OT or physical equipment environment, as Camilo was mentioning, makes it very, very easy for those kind of like cyberattacks to cascade very fast, right? Somebody targeting one system can shut down a company right away. So that is, like, one of the biggest challenges in the sense that the level of coupling makes it very, very hard to basically stop, even if, of course, there is a lot of, you know, network segmentation that can be done, and cybersecurity teams have been doing that for years, but it's much harder than before. And the other one I would say is complexity. That's usually when you look at catastrophic disasters, being able to explain what went wrong, right? And how the system is gonna behave. For instance, in space, there's so many things that we don't even know, right? How reactions are gonna happen, things like that. It's almost impossible to predict sometimes. With cyber nowadays, for instance, because we have a lot of AI that is being deployed in modern supply chain technologies, all across the board, that is gonna become also an issue, right? We don't know a lot of the times how these systems are gonna behave, right? I mean, that's part of artificial intelligence that is nondeterministic. So that's gonna become, I think, in the next, you know, five to 10 years, another big challenge in terms of how do we deal with that uncertainty in terms of you have a bunch of different, you know, AI tools deployed, maybe AI agents in your supply chain, and all of them start maybe misbehaving, but you don't understand it, right? And that can scale and cascade pretty fast. - Yeah, and also I would say that another important challenge is talent, right? So, there is a significant shortage of cybersecurity professionals worldwide. Some organizations estimate that it's around 3 million, like, people, others around 5 million people, and, you know, it's a matter of, in this case, not just about cyber, but having, you know, the right combination of skills. There is a question, like, we want to also address with the lab, is how to upskill, you know, supply chainers that have no clue about these threats. - Yeah, and we already have an idea of a new position that we are gonna be probably, you know, naming. What is it? - Sorry. - Supply chain CISO. Probably we are gonna have dedicated CISOs per function in the organization, right? Most of the companies, they do have a global chief information security officer, but we know certain industries, for instance, the financial institutions, which are usually, like, ahead when it comes to cyber because of, you know, what it entails to protect banks. - Yeah. - They do have different, like, CISO positions for, like, specific verticals of the company, right? One might be for commercial banking, one might be for, like, internal, you know, IT systems. So, to be honest, considering the deployment of technology all across the world, and specifically in supply chain operations, we're gonna need supply chain CISOs soon. So specific people that look at how cyberattacks can impact physical operations in companies. - So we have a model in banking, and it should cascade throughout, right, to protect that. - The benchmarking part is, and it's great that you say that, Mackenzie, 'cause one of the biggest things I think that we can do from the research perspective, like, at MIT, is that. Is that kinda, like, you know, cross-pollination of knowledge from different industries. So benchmarking, okay, what industries maybe are ahead when it comes to cyber? Where industries are ahead when it comes to supply chain? Bring all that knowledge together. - Certainly. In terms of supply chain, cybersecurity is not just about technology. It's about people, processes, leadership. So those new positions, those new roles, you know, for these challenges of the 21st century, right? So imagine, you know, that, like, how a company manages a catastrophic event or this kind of crisis. Who makes the decisions, right? We were, you know, discussing this with leaders, you know, from different companies in our roundtable that we conducted here at MIT, and then the question is, like, who communicates with suppliers and customers? Which systems should be isolated? Which operations can actually continue operating normally, or manually, right? So the companies should train people on how to deal with these circumstances, you know, for the decision-making process. - Well, that's an interesting question, because we're having a push for companies to use more technology, be more digital, and yet if there is this kind of catastrophe, of course, that knocks those digital systems out, you have to have people who know how to do things manually. How do you all suggest companies finding that balance? - I mean, given that they already experienced these disruptions. So in our interviews, we have tried to understand what they do, you know, when this happens. So as I described before, we have the silos. So the IT team is trying to recover the systems, while the supply chainers are trying to, you know, continue operations. So this is, of course, contingent to the type of company. And so, but some of them, like, "Well, we started creating our new systems," like, communication channels like WhatsApp. I don't know, getting information from some spreadsheets, phone calls. You know, like, going back to 1980. - Back to spreadsheets and phone calls. Yes. - Right? It's like turning the switch off and operating manually. Because again, even if the IT team says, like, "Oh, we got, like, the systems restored," and whatever, operations is like, "Can we trust that system? Is the malicious actor still there?" - Yeah, and there are a lot of trade-offs there. 'Cause for instance, okay, we have the IT system restored or the main IT system, but what about all the shadow AI, you know, tool that everybody's using already but was not official? Is that actually restored? Or a lot of the times, all the Excel macros that people were using for the daily work, and there is a real example of a big company that suffered that, right? Not too long ago. Like, their systems went completely down. They have one of the best cybersecurity teams worldwide, I would say, and they recovered the main systems pretty fast, in less than a day. what they didn't know is that people were actually using hundreds of other systems in the back, like, and those, of course, couldn't be restored officially 'cause nobody knew about them but the specific teams that were using them. So really getting back to the level of productivity that they had before restoring all those shadow systems, that usually takes a lot longer. And the same can happen thinking the other way around. So you might be able to restore, for instance, the operational technology and say, "Okay, the equipment is ready." 'Cause you need to make sure that your systems are secure, that the attackers are gone and not again in the system. So you need to do a lot of forensics for that to be, let's say, ready to go. But still, a lot of the data is probably not synchronized anymore, right? After you went through an attack, you probably had to do a bunch of, you know, cleaning. A lot of the data that was coming in, as you shut down your systems, it stopped, so probably you're gonna have a gap in between. And a lot of the new technologies, they require, like, updated data and the pretty synchronized data to keep moving, right? So even if the robot or the data specific technology is ready to go, a lot of this data synchronization takes a long time to really get back to it. So, and that, again, this example, the OT, the equipment, the physical equipment was recovered in less than eight hours. Now, really to get back to the productivity levels and get back to have the perfect data to keep operating took them days. - Wow. - Yeah. - And for leaders listening who are wanting to improve the cybersecurity in their organizations, of course, everything is a trade-off, what people are, you know, choosing to focus on. What trade-offs must leaders make when looking at strengthening cybersecurity? - So I think we've mentioned a few already. This one, for instance, about the, you know, how fast you wanna recover and how many forensics you wanna do, that's a big one. You can try to recover very fast and basically reset all your systems without looking at what happened. And basically, the idea of doing that is to try to maximize business continuity, right? "My plant needs to keep operating. I shut down everything, I reset it. In one hour, we are good to go." The issue with that is that probably you had no time to do any forensics, so the attackers might still be, - How much can you trust it, yeah? - Exactly. Like, are your systems compromised? Probably you don't know. So the chances of having another attack, like, soon are much higher, of course. So that's one. We talked today a lot about the increasing in the amount of suppliers, mainly in the technology space for supply chains, right? So usually in traditional risk management in supply chain, we were always taught that, you know, you diversify suppliers to basically de-risk. 'Cause if something bad happens to one of them, for whatever reason, you just, you know, swap it and you go to the next. In cyber, I mean, that still, you know, remains true, when it comes to the physical products. But the issue is that as you basically expand your number of tier one suppliers, you also expand, you know, your attack surface. And it's not only that direct supplier, but also all the tier two, three, four suppliers that come with that supplier that are gonna be part of your new attack surface. 'Cause a cyberattack basically cascades pretty fast through the supply chain. So that's a new important trade-off when it comes to how big is gonna be your supplier base. 'Cause you're de-risking from one side, but at the same time, you're increasing your attack surface from a cybersecurity perspective. - And going back to the role of AI in cybersecurity, how is AI changing supply chain cybersecurity, both as a threat and as a defense? - Well, companies can certainly leverage AI for defense. So you know, they have these mature cybersecurity teams that can, you know, have access to certain models, improve or patch certain system scripts, so on and so forth. But on the other hand, you have, like, these malicious actors that can also, you know, use these new models to potentially, you know, attack certain companies. Now, I believe that something that's gonna happen is that these, let's say, hackers are not gonna be able to really disrupt some large firms by attacking them directly. So they are going to go after small and medium-sized companies that do not have the capabilities, the resources, the muscle to really prepare, you know, and defend from these attackers. They are not even aware of that. Small, medium-sized companies that, with vibe coding, are developing certain tools, you know, that's great, you know, to be, like, more efficient. But at the end of the day, they do not know what's in the back end, and hackers can exploit that. - And we have seen that trend already, right? When we were talking about the technology providers being the ones hit, 'cause usually those tech providers of the main firms, they're actually smaller in size. They're SMEs, lower cybersecurity capabilities. And of course, again, yeah, with tech providers, it happens that the return of investment of the attacker is the highest 'cause they can have this cascading effect. Actually, in the roundtable that we hosted, we did see a big difference, right? In certain companies being very open about using AI and deploying it, others being more reluctant for now, a little bit, yeah, hesitant. But still, you know, those large companies, probably they are gonna have, you know, clear procedures of, okay, how do you take an AI tool that someone is building to really be, like in the production stage and ready to deploy. But SMEs are gonna be deploying those, like, super fast 'cause of course, when you're a small company, all that matters is business, and you don't have money usually to pay for cybersecurity. And I think a good example of that actually is that, for instance, for the defense industrial base of the U.S., of course, you have the big primes, right? Lockheed, Northrop, and others. But there is a huge amount of small, medium-sized suppliers that provide to those, right? Early this year, and throughout the year, they're having a lot of conversations in terms of increasing the level of cybersecurity for a lot of those small and medium-sized firms. And actually, the government wanted to pass a new regulation called the Cybersecurity Maturity Model, and they had to basically stop it 'cause the level of cybersecurity requirements that it would put into these small suppliers, they just couldn't handle it from a cost perspective. It was way too expensive for them to really, you know, get to the level of cybersecurity that supposed to be required. I wanna go back now to the idea of why we wanna create a SCOUT, right? The main target of SCOUT, of this new tool that is just, the beginning of that right now, 'cause we know it's not gonna be easy to build and we are working already with companies to basically do it together, is that it's gonna impact mainly small and medium-sized firms, right? A tool that allows you to map the vulnerabilities to basically be able to see where you should act and how you should act to improve your cybersecurity capabilities in the supply chain, how you need to mitigate certain risks, how you need to prepare for recovery in case of an attack. - There's been a lot of conversation and emphasis on agentic AI. As AI agents take on more work, what new risks do they create, and what guardrails are needed? - 2026 has been the year of agentic AI, and- - And seeing in the news recently, some agents misbehaving, - Yeah, are going rogue. - Yeah. - Going rogue. - They're going rogue. AI is nondeterministic. You can tell it whatever you want, but sometimes it's not gonna do exactly what's in the prompt 'cause, of course, you can only prompt so much. I mean, it's basically impossible to give infinite instructions, right, and be super precise, and that's actually one of the risks, right? AI still does not distinguish very well between instructions and anything that comes after that, like any input, and that's actually one of the biggest risks with any kind of AI tool and, of course, agents, right? Like prompt injection. - Sounds like at the toddler phase, right? - Yeah, yeah, of course. Like something that, you know, gets embedded into your AI model, from a malicious actor, which I think most of the cases that we've seen so far are not related to that, are related to just the AI misbehaving. But, of course, we are definitely gonna see more and more, cyberattacks targeting AI agents in the near future 'cause they're gonna be everywhere, right? So we are gonna be seeing probably AI fighting AI more and more. But the other day also in conversations with some of the companies that are our partners, some of them are deploying AI as a decoy, right? They're actually basically duplicating all their main kinda like infrastructure or some of it to basically show as a decoy to any kind of attacker. So instead of the attack going straight to your organization, like IT infrastructure, it goes to this AI, kinda like fake company. - Give them an easy target. - Yeah, giving them an easy target. - And it crumbles. - And because of most of the attacks are now AI-enabled, they're also spending a lot of money to perform the attack. So basically what companies can do is with these AI fakes, they actually make the attacker spend a bunch of time just going through the system, getting lost, at the same time that they're spending a lot of money, right, because they're paying for tokens to really perform those attacks. So yeah, it's becoming a pretty interesting, you know, battle between AIs, nowadays, yeah. - For companies listening, what would you all suggest that they can do to improve their resilience? 'Cause it can feel like a tall order to improve your cybersecurity. What should they do first? - Three things. The first one, people. The second one, people. And the third one, guess what? People. So yeah, technology matters, but people remain one of the greatest sources of both vulnerability and resilience. As a company, you can invest in sophisticated cybersecurity tools, but if employees are not aware of the risks, if leadership does not know how to respond during a crisis, or if different functions didn't know how to work together, the organization can still be extremely vulnerable, right? So I would say cybersecurity preparedness is ultimately about people, culture, and leadership, not only technology. - Yeah, when we started working also in supply chain risk management and then, of course, we got into cyber, we used to look at three main ways of preparing for any kind of disruption, right? In risk management, you look at prevention, mitigation, and recovery. Of course, in cyber, the detection part is a big part of it too. But I think, for most companies, to mapping those dependencies across, you know, the IT and the physical operation side is gonna be huge. Prevention, still pretty much, I would say, 90% of the work is gonna be done by the cybersecurity teams and IT teams. But having the supply chain people and operations people at least being aware of it and getting involved in the conversations is gonna be super important. And the same, that goes into the mitigation and recovery part 'cause that's where most of the supply chain people, operations people need to play a much bigger role than they have in the past. So for instance, in cyber, it's very typical to do tabletop exercises, right? To basically model how an attack, you know, is happening, and once it has hit the organization, how we are gonna respond to it. Usually, they're pretty strategic, so most of the times, they remain at the high level of the organization, right? C-suite, VP level, and so on. But it's critical now to start bringing operations people into them. It's not just a data breach, reputational issue. No, this attack just stopped your manufacturing plant from producing. So really having like the people from the field in those exercises and going from, again, the prevention, mitigation, and to the recovery part of it altogether is gonna be critical. It's super hard 'cause also when you think about the business impact of a cyberattack, that might change like from one day to another, right? Most of the attacks don't happen during valleys of demand, for instance. They happen during peak season, like, you know, Thanksgiving, Christmas is coming, all the retailers are getting hit. 'Cause, of course, for the attackers, I mean, they know like if they're asking for a ransom, oh, they're gonna pay 'cause they have all their, you know, Thanksgiving shipments ready to go, and they need to, right? How did you actually model the risk in a different way when it comes to that, right? Probably you're not gonna have the same kind of recovery plan for a normal day than, you know, the week of Thanksgiving or two weeks before Thanksgiving when you're basically shipping out everything, Black Friday and so on, right? So those kind of like dynamic business continuity plans, for instance, is something that, you know, the operations and supply chain teams need to be involved in the conversations early on. - And I will add to that that one practical recommendation, you know, for companies would be identify the operations you cannot afford to lose. Understand what they depend on, and prepare your people to operate when those dependencies fail. That is critical. - You all mentioned it's hard to look ahead, because there's so much changing in cybersecurity, but if you look at the next five to 10 years, how do you all think supply chain cybersecurity may evolve in that time? - Oh, well, we are developing a tool for that. It's called MIT RADAR, and it's essentially a tool that is helping us to identify emerging cybersecurity risks before they become a major supply chain threat. The idea is to respect the different domains, you know, like physical operations and IT and others, even like, you know, signals like quantum computing, how these, you know, technologies are gonna play a critical role in the future, how companies should prepare their systems, organization to those trends. It's gonna happen, right? So again, like the idea is to combine the learnings from TRACE and start like, you know, trying to anticipate the future with RADAR to shape the present, the decisions today. - And you all have spoken a lot about, you know, what the lab is working on in terms of tools, but in terms of everything else, what is the lab doing now? What are the priorities, maybe things that you haven't covered yet? - Hmm. - Yeah, to sum up the, maybe the tools part, 'cause it's something that we're very proud of. TRACE, looking at the past or mapping all the cyberattacks worldwide that are having an impact on physical operations. RADAR, looking at the future, so looking at all the emerging technologies that are coming to supply chain and basically crossing all that information with current vulnerabilities that we're seeing in the space already to tell a company, "Hey, if you're gonna be having humanoids in your warehouse tomorrow, how are you gonna have to think about cyber from that perspective?" And of course this is gonna be huge 'cause we are gonna have more and more autonomous systems, not only agentic AI on the digital world, but also, you know, autonomous trucks coming to the roads, humanoids coming to manufacturing plants, warehouses, so. And then SCOUT, that basically puts all together, kind of like takes information from those two and becomes a vulnerability assessment tool, dynamic tool, for a company to be able to actually just improve their cybersecurity capabilities when it comes to supply chain. Those are definitely, I would say, the main pillars of what we are working on already. And of course, then we are doing interviews. We call that the impact study with a lot of the companies that have suffered some of these large cyberattacks worldwide and events like the roundtables. So the idea is to bring experts together and to learn from all of them, right? To see what's going on in this space. A lot of these cyberattacks are not being disclosed openly, and I think every time I actually teach a class, I always ask, right, like, "Who has, you know, been impacted by a cyberattack?" I think four or five years ago when I started having kind of like these ideas in mind, like there were basically a couple people raising their hands. In the last class, I taught, like 70% of the class were already like, "Eh, me." So that's definitely, you know, a big change. So the third pillar at MIT, and specifically at CTL, right, is educational piece. So I've been teaching already for a while, on cyber, but we are already thinking ahead, and considering a course on supply chain cybersecurity, a specific course. We don't have dates yet, but we think 2027 is pretty likely for us to have that ready, and we're super excited for it. - So those are all the questions I have, but anything else that you all wanna add? So great to have you both on the podcast. - Thank you. Sure, for our listeners, we are here to help you, so please feel free to reach out, you know, to explore our website, our tools. We work every day to help you, to help companies. It may sound a little bit romantic, but at the end of the day, we are doing this for a better world. - From my side, big thank you too. It was a pleasure. Our goal, again, is to protect physical supply chains from cybercrime. That's why we created this new lab. Any company that wants to talk to us, we wanna learn about what they're going through, and hopefully then we can bring that back to the industry and make cybersecurity for supply chains, you know, just better for everybody out there. - That wraps up this episode of "Supply Chain Frontiers." A big thank you to Miguel and Camilo for sharing their expertise and insights into the current state of supply chain cybersecurity. To learn more about the lab, visit it on the research page of ctl.mit.edu. "Supply Chain Frontiers" is recorded on the MIT campus in Cambridge, Massachusetts. Our sound editor is at David Benjamin Sound, and our audio engineer today is Kurt Schneider of MIT Audiovisual Services. Our producer is myself, Mackenzie Berry. Be sure to check out previous episodes of "Supply Chain Frontiers" at ctl.mit.edu/podcast or search for us on your preferred podcast platform. I'm Mackenzie Berry. Thanks for listening, and we'll catch you next time on "Supply Chain Frontiers."