Drawing a digital line for geofencing.
Dave Bittner: Hello, everyone. And welcome to Caveat, N2K Cyberwire's Privacy Surveillance Law and Policy podcast. I'm Dave Bittner. And joining me is my cohost, Ben Yelin, from the University of Maryland Center for Cyber Health and Hazard Strategies. Hey there, Ben.
Ben Yelin: Hello, Dave.
Dave Bittner: It's a big show today, Ben. We're discussing the Supreme Court's landmark decision on geofence warrants and, later in the show, former congressman and current president of Americans for Responsible Innovation, Brad Carson. He's here to discuss the US-China trade relations and how they could shape the future of AI development. While this show covers legal topics, and Ben is a lawyer, the views expressed do not constitute legal advice. For official legal advice on any of the topics we cover, please contact your attorney. All right, Ben. I know you are just -- have been so excited. You've been sitting on the edge of your seat, waiting to see how this Supreme Court ruling was going to come down; and it has happened.
Ben Yelin: It did. It's a huge day for us. I couldn't believe it. I knew it was going to come down either the day before we recorded our pod or the day we recorded our pod. Lucky for us, it was the day before.
Dave Bittner: Yeah.
Ben Yelin: So I've had about 24 hours to digest this case. It's a landmark for us. It's surprising in a lot of ways. And it's a landmark case for digital privacy, and I think it's a case we'll be talking about for a long time. I know I'm going to have to change a lot of my syllabi for the courses I teach to include many aspects of this case. So I'm thrilled to be here with you and ready to talk about it.
Dave Bittner: All right. Well, let's start with just a quick review of what got us here. Give us a little of the background.
Ben Yelin: So this is a case called Chatrie v. the United States. There was a bank robbery in the state of Virginia, and law enforcement was looking for leads. And they went to Google and said, Can you tell us which devices were in the area of this bank robbery when it occurred? That's called a geofence warrant. So you have this temporal restriction and then this geographic restriction. There's a three-step process to these warrants when they existed. The first is getting unique identifiers for all of the devices that are there. And, once you get those, law enforcement can kind of narrow it down and pick out a number of devices that they think might be suspicious for one reason or another. And then, in part two of the warrant, they can go back an hour before the crime took place, look an hour after for that limited subset of devices to see where they were and what they were doing. And then, once they use that as an investigative tool, step three of the warrant is demystifying the data.
Dave Bittner: Deanonymize.
Ben Yelin: Deanonymizing.
Dave Bittner: Yeah.
Ben Yelin: I have so many smart thoughts about this case, and I couldn't come up with the obvious word there.
Dave Bittner: Right, right.
Ben Yelin: Deanonymizing the data, they did that. They figured out it was this -- likely this guy Chatrie. He was arrested and prosecuted. He seeks to suppress the evidence, saying that this geofence warrant was an unconstitutional search under the Fourth Amendment. He is able to appeal his conviction to the Fourth Circuit Court of Appeals. They have a very divided opinion that sustains his conviction, but it's very unclear where they come down on the constitutional issues. There are a lot of constitutional issues at play in this case, and so the Supreme Court wisely takes this up. There are two threshold questions in this case. The first, and the one that the majority opinion here addressed, is whether this counts as a search for Fourth Amendment purposes. If there is no search, then we don't have to go into the complicated analysis of whether this was a constitutionally valid warrant. The majority opinion here, spoiler alert, held that there is a search. So that's the major finding of this case, and we'll get into their rationale in a moment. The second portion of this case that is relevant, the second major question is whether this was a constitutional search, particularly whether the warrant here is based on probable cause and is properly particularized, which is a requirement in Fourth Amendment jurisprudence. And the majority here did not reach that second question. They remanded the case back to the Fourth Circuit and said, Fourth Circuit, you figure out the details of this warrant. Whether it was a defective warrant or not, we're not going to get there. All we are deciding today is that this is a Fourth Amendment search. Therefore, the warrant requirement applies. So it is narrow in that respect. It only bears -- the majority opinion only bears on that first question.
Dave Bittner: Okay.
Ben Yelin: So it is -- I would characterize this more as like a 5-3-1 decision.
Dave Bittner: Really.
Ben Yelin: Yes. So we have our --
Dave Bittner: I just -- wait. I saw someone this week who said that they should have -- someone should print up T-shirts that say like Supreme Court 6 to 3, right, because, like, it's -- it's the number -- it's so representative of where we are for this time, right?
Ben Yelin: Yes, it is. And I guess you could technically call this a 6-3 decision because there are six justices who agree that this was a constitutional search.
Dave Bittner: Okay.
Ben Yelin: But one of them is just coming at it from a completely different direction that I think his decision is so far afield from what the majority actually decided -- and trust me. We'll get to all of this. I'm going to try and contain myself here. So the majority opinion was drafted by Justice Kagan, one of the liberal justices. She's joined by the other two liberal justices, Justices Jackson and Sotomayor, along with Chief Justice John Roberts and our old friend Brett Kavanaugh, who -- he was not on the court when the court decided United States v. Carpenter in 2018. So it was kind of unclear where he would fall on these Fourth Amendment digital privacy issues. And so the fact that he's in the majority opinion here was particularly illuminating. So let's get into the substance.
Dave Bittner: Wait. Who's the one?
Ben Yelin: The one is Justice Gorsuch.
Dave Bittner: Okay.
Ben Yelin: He has an entirely different theory on the Fourth Amendment and how we should look at Fourth Amendment cases in light of new technology. And he had that opinion when he was on the court. It was actually his first year on the Supreme Court when he was a dissenting opinion. He wrote his own dissenting opinion in the Carpenter decision. It's a property-based approach that doesn't follow what the rest of the Supreme Court has generally done with Fourth Amendment jurisprudence on new tech, on new technology. When we have these kind of untested modern technologies that present these novel questions, the Supreme Court has gone about an analysis in a certain way. And Justice Gorsuch rejects that method of analysis. So he's -- he's the one. And then we have a dissent. And hot take: There are a lot of parts of the dissent I find quite compelling, even though my biases, I'm pro digital privacy; so I'm happy with the outcome of this case. I think, for digital privacy rights, this is a groundbreaking decision. But there's a lot in the dissenting opinion written by Justice Alito and joined in full by Justice Thomas and in most parts by Justice Barrett that I think are quite compelling. So let's start with the majority opinion by Justice Kagan.
Dave Bittner: Okay.
Ben Yelin: Basically, the Supreme Court has settled on a test for what is -- what qualifies as a Fourth Amendment search, and it's this framework around a reasonable expectation of privacy. So this comes from a decision called Katz v. the United States from 1967. One of the concurring opinions in that case from Justice Harlan came up with this two-part test that, if a person has a subjective expectation of privacy that society is willing to recognize as reasonable, then you have a Fourth Amendment event, so to speak, or a Fourth Amendment search; and the warrant requirement applies. So what the majority is trying to argue here is that this search, this geofence warrant was a violation of Chatrie's reasonable expectation of privacy, that people have a reasonable expectation of privacy in their location data. So what do we mean by location data? It's not CSLI. That was Carpenter. Carpenter was about historical cell site location information. So I turn on my phone, and it pings a cell phone tower. That's what Carpenter was about. And that can track a whole of a person's movement over a long period of time as you go from cell phone tower to cell phone tower. This is different. So enabling location services is something that we all do without really thinking about it because it improves the functionality of apps, especially Google Apps. Because I'm a user of Google Maps, and if I didn't enable my location when I downloaded Google Maps, it would not be very effective, Dave.
Dave Bittner: Right. Many of the -- many of the apps that we use depend on it.
Ben Yelin: Absolutely. If you have an Android device, you'll actually get a warning if you try and turn off the location services that says explicitly, If you turn this off, you're going to lose a lot of functionality on -- on your applications. You don't get that explicit warning on Apple, but it becomes clear after a short period of time that, like, to make the best use of your applications, to find the nearest fast food joint or to see what the weather is going to be like in a particular area, you should be turning on your location services. As opposed to CSLI, the user has to kind of make -- do something affirmative here. They have to take an affirmative step, which is to enable location services, location tracking in a way that just doesn't exist with CSLI. I think that's an important distinction. The court here seemed to think that there really wasn't that much of a meaningful difference. It's not -- just as with CSLI, it's not really something that's voluntary because everybody needs a cell phone this day and age. And so turning on your phone and having it ping cell phone towers isn't any sort of voluntary action. I think the Supreme Court sees this similarly here. If you want to enjoy the full use of your cell phone, of your smartphone, you kind of have to turn on location services.
Dave Bittner: Yeah.
Ben Yelin: The counterargument to that is -- and this is something that's mentioned in the case -- a majority of iPhone users, at least, do not turn on Google location tracking. So there certainly is a way to have a smartphone without using this particular technology; and, therefore, I think there is an element of voluntariness to the transaction here where I certainly realize -- maybe, you know, some people might not realize this. But I and I think other users of these devices realize that, when you're opting into location services, like, you are going to be sharing some potentially private and valuable data with Google.
Dave Bittner: Yeah.
Ben Yelin: The reason this is relevant is because of something called the third-party doctrine. So, historically, the Supreme Court has said that people do not have that reasonable expectation of privacy in data that they've handed over to a third party. So the original cases for this were things like bank records and the phone numbers that you dialed through a pen register. That had been kind of a universal rule for all third-party records cases until Carpenter. Carpenter was the first one when we had an instance of Mr. Carpenter voluntarily, by turning on his phone, sharing information with a third party, his cell phone provider, but the Supreme Court saying this isn't a third-party doctrine case because we're talking about more than phone numbers and -- phone numbers dialed and bank records here. We're talking about the depth of the information, the breadth of it, and its involuntary nature. That makes this fundamentally different. What the court here is doing in Chatrie, what the majority is doing is saying, what we said in Carpenter, we're doubling down on that for Chatrie. That is just -- it is just as true with location services as it was with historical cell site location information, that the depth of the information being revealed is extremely significant. The breadth matters. And the lack of voluntariness, the fact that cell phones are so pervasive, they're a part of our modern life and that we would use the crucial function -- lose the crucial functionality of cell phones if we had to disable these services. The fact that we have this depth, breadth, and involuntary nature of the data collection means that people actually do have a reasonable expectation of privacy in their location data. Some of the best evidence of that is we at least have the illusion that we control that data. It's ours. I can look on my device, go through the depths of the Google Map application, and see every location I've been at over a period of time. And that could reveal a lot of private information about me; could reveal my political and religious associations and a whole bunch of different other things. So the Supreme Court is extending the logic of Carpenter to this case and saying everything we said about cell site location information is just as true, if not more true as it relates to location services. So that is the crux of the majority opinion. This is a search because we do have a reasonable expectation of privacy in this data, and the third-party doctrine does not apply because this is an add-on to Carpenter. This is a case where we have a significant amount of information being turned over, and it's being done in a way that's not exactly voluntary because of the nature of this technology.
Dave Bittner: So help me understand here. Is the -- was the Supreme Court's ruling about location services writ large or geofencing specifically?
Ben Yelin: It was more location services writ large.
Dave Bittner: Okay.
Ben Yelin: And I think they make that distinction very clear in the case that geofence warrants are different in a number of ways. One of them is they're usually for a short period of time. So, as it relates to this particular incident at the bank, they were only searching for about two hours worth of data; and then it's only for a limited geographic area. The Supreme Court's majority here is saying that those distinctions, the temporal distinction and the geographic distinction, aren't significant from a constitutional perspective because it would be an impossible line drawing problem to be like, okay. Well, two hours, you know, maybe you should have a reasonable expectation of privacy, about two hours worth of your location services data. But three hours, four hours, does that make a difference? You can see why that would be a very difficult line to draw. So what the majority says here is this is not about the particulars of the geofence warrant. It's about the fact that this is based on location services, and that's the data that we're concerned with here. We are analyzing this based on the data that is being revealed to the third party and thus through the warrant process to the government when it has issued this warrant to Google.
Dave Bittner: Now, let me ask you this. If I'm the CEO of Flock, who's the license plate reader company, am I nervous?
Ben Yelin: I think so.
Dave Bittner: Yeah? Okay.
Ben Yelin: So one of the points of emphasis here is that you can track a whole of a person's movement through their location services. I think there are meaningful differences between tracking a whole of a person's movements through location services versus what Flock does, which is a license plate reader.
Dave Bittner: Right.
Ben Yelin: Let me just ask you this question: Would you rather share your past location solely by showing people where your license plate was read or showing where Google pinged your location over an equivalent period of time?
Dave Bittner: Yeah. I mean, certainly the license plate is going to be more diffuse. It's not going to show as many locations. But still, you know, you could learn a lot about -- by tracking someone's travels, as we've said, you can -- you could learn a lot. And I guess the -- a -- to me a decent question is how granular do we have to get with these license plate readers -- because they're popping up everywhere, right -- to be a problem?
Ben Yelin: I think -- again, I think the distinction here matters. One of the things they talked about is how good location services is on multiple dimensions. So it not only can identify where a person is within, like, a very limited number of square feet. Like, you can get down to something that's very granular.
Dave Bittner: Which side of the street you're on. Yeah.
Ben Yelin: Exactly. But it also -- there's the second dimension, which is you could be in a 100-story skyscraper, and it could tell you generally which floor a person was on. And you can see how that would become very relevant in an investigation.
Dave Bittner: Right.
Ben Yelin: Like, let's say there was a bank robbery on the first floor, but somebody lived in an apartment on the 30th floor. You'd really want to have that distinction, and location services gives you that distinction. So it just is much more granular.
Dave Bittner: Right. Yeah, yeah. I'm thinking there's a hair salon on the 10th floor and an abortion clinic on the 5th.
Ben Yelin: It's the craziest building in America.
Dave Bittner: Right, right.
Ben Yelin: I think the reason that the Supreme Court may not extend this type of reasoning to Flock -- and, again, we don't know the answer to this question because we haven't seen a case -- is just how granular the data is here.
Dave Bittner: Okay.
Ben Yelin: How exacting it is. That gets to how revealing the data is, and that's why we need Fourth Amendment protection to maintain this balance between law enforcement's capabilities and people's freedom in privacy from the government's peering eyes. And I think it is the particularly revealing and accurate and granular data that we're talking about here that motivated the majority to come to this conclusion.
Dave Bittner: Well, let's talk about the dissent. You said there are some things that you found compelling in there.
Ben Yelin: I did.
Dave Bittner: Yeah.
Ben Yelin: So the dissent is drafted by Justice Alito, and he does get into the merits of the case. But, before he does that, he talks about why he doesn't think this case should have ever come to the Supreme Court. The Supreme Court, per Article III of the Constitution, is only supposed to consider active cases and controversies where there's a real stake in the outcome. Justice Alito argues here, I think compellingly, that there isn't actually any doubt about the ultimate outcome of this case, vis-à-vis Chatrie himself. So Mr. Chatrie was convicted. The Fourth Circuit said, you know, maybe the warrant was defective. Maybe this was a Fourth Amendment search, but Chatrie is not getting out of prison because of what's called the good faith exception. If law enforcement are relying on a reasonable interpretation of the law at the time of the investigation, then there's a presumption of good faith. The exclusionary rule doesn't apply, and the person is going to stay confined behind bars. None of the parties in this case argued at any time in any sort of compelling way that ultimately the good faith exception would not apply. In other words, everybody sort of agrees that, no matter what the Supreme Court decides here, Chatrie himself is not getting out of prison because it's pretty obvious that he robbed a bank; and it's also obvious that law enforcement were acting in good faith when they conducted this geofence warrant.
Dave Bittner: Didn't he, like, plea out except for this one issue, something like that?
Ben Yelin: Well, it's not really -- I mean, it's -- he is still a criminal convict, and his -- his overall guilt is not going to be reversed --
Dave Bittner: Okay.
Ben Yelin: -- as Justice Alito claims as a result of this decision.
Dave Bittner: Okay.
Ben Yelin: And that's ultimately what matters.
Dave Bittner: Yeah.
Ben Yelin: There is no way he is going to get out of his criminal culpability because it's very obvious that law enforcement acted in good faith. I think that's frankly a compelling argument. Like, the Supreme Court should only hear active cases and controversies. They should not be issuing what are referred to as advisory opinions because that's not their proper constitutional role. The counterargument to this, which I also think is compelling, is we would never reach the merits on a case about a new technology if for every single one of these cases we were relying on the good faith exception, because the only way law enforcement can have constructive knowledge that a particular piece of technology or technological surveillance tool, the only way law enforcement can realize that that poses constitutional problems is if a court tells them so. And the only way a court is going to tell them so is if they get to the merits of one of these cases. Does that make sense?
Dave Bittner: Yeah, yeah.
Ben Yelin: So I think Alito's view here might be technically and doctrinally accurate, but I think it would preclude us from getting clarity on these critical constitutional Fourth Amendment issues about whether this counts as a Fourth Amendment search. The other thing Alito mentions, which is also compelling to me, is that there's a mootness element to this. We've talked about it. Google stopped housing the capability to fulfill these geofence warrants sometime in the summer of 2025. All this data is now held on the device and is not uploaded to the cloud where Google can access it. So, yes; it might apply in this case because Chatrie robbed a bank prior to Google making that decision. But this isn't something that's going to have significant implications going forward because Google, which corners the market on this stuff, doesn't maintain the data to fulfill these geofence warrants. So, to me, that's also compelling. Like, courts should limit themselves in the cases that they decide. And here we have a case that might not be justiciable, because, A, it's moot; and, B, it's not an active case or controversy. So I certainly find something compelling about that.
Dave Bittner: But there are other apps on our phones to track our location, like Facebook notoriously --
Ben Yelin: Yeah. I mean, that's what the majority opinion said is that, yes, Google is a significant portion of the market; but, A, there are all these other apps owned by other companies that might be able to fulfill geofence warrants; and, B, Google could get a new CEO tomorrow who decides, You know what? We want to cooperate more with law enforcement. It's good for our bottom line. Let's restore the capability to execute these geofence warrants. And that's, I think, a reasonable argument as well. But I certainly don't think -- at least as it comes to those two points, I don't think Alito was coming at this from nowhere. I think those justiciability questions are critical here, and his arguments were particularly compelling.
Dave Bittner: To what degree was this ruling surprising?
Ben Yelin: It was surprising to me in the -- in kind of how far the majority opinion went in extending Carpenter. There was a lack of clarity after Carpenter because that case was so limited to the facts before the court about historical cell site location information. There was this bubbling question about how far that would extend in other cases. And so much of the emphasis in the majority opinion here is about the unique nature of smartphones, that there's something unique about smartphones in that we have to have them. We all have them. They're basically an appendage. We need to use them to fulfill the duties of everyday life. And, when it comes to the core functions of smartphones, we should have a reasonable expectation of privacy that, you know, given how much data we keep on these phones, the government isn't going to come prying into our private affairs. I didn't think they were going to go that far and say not only does that logic apply to cell site location information, it also applies to location services. And the reason it applies to location services is because the smartphone itself is so significant. That has downstream impacts that I think we're going to be seeing for a long time. This is something that Alito warned about in his dissent. There are a lot of --
Dave Bittner: Yeah. That's going to be my next question.
Ben Yelin: Yeah. I mean, this is the slippery slope here --
Dave Bittner: Okay.
Ben Yelin: -- is Justice Kagan emphasized the normal features of smartphones. One of the normal features of smartphones is making phone calls. I know we don't do it very much.
Dave Bittner: Right.
Ben Yelin: But we've had this 50-year precedent that you don't have a reasonable expectation of privacy in the phone numbers that you dial. Is -- that's obviously true for landline phones. Is that still true for smartphones? Because what Kagan is saying here is smartphones are different. They're unique. There's so much private information on them. Is there going to be a different judicial standard based on whether we made a phone call on a landline or a cell phone? And there are a million other questions like that, that can come up. There are a lot of normal functions of our smartphones. You know, can the government track what we purchase on Amazon? That seemingly would be another third-party records case. We know or have constructive knowledge that Amazon keeps a list of all the things that we've purchased. But, given that we did so with our smartphone and given what Justice Kagan is saying about smartphones in her opinion here, is that information now constitutionally protected? And is the government going to have to get a warrant when it wants to search our Amazon records? And are those records any less private than the bank records that were an issue in previous third-party records cases like United States versus Miller? And we don't know the answer to those questions. So I'm surprised in how broad it was, and I think it's going to take us a while to understand how lower courts are going to apply this reasoning that's so crystal clear on the preeminence of smartphones and their uses and their providence in our daily lives. I'm very curious how that's going to be applied in all of the circumstances that come across our lower courts.
Dave Bittner: So just so I'm clear here, let's go back to Facebook as an example. So we know that, for example, Facebook collects our location data. And they use that for advertising; they use that for suggesting friends and things that we may be interested in. So this case -- is my understanding correct that this case really does nothing to address the collection of that data? What this case has to do with is whether -- is what the government needs to do to get access to that data.
Ben Yelin: That's exactly right.
Dave Bittner: Okay.
Ben Yelin: So there might be state laws or at some point federal laws that restrict the ability of companies to collect this type of data. All of that is still permissible, but what this case is about is whether the government needs a warrant to obtain this privileged digital information. I'm using that as kind of a term of art but the information that we're talking about here, which is location services and any other, the way Kagan puts it, like, normal data generated from the -- from the use of smartphones, so that is the central issue here. And because the Supreme Court decided that this counts as a constitutional search, the answer is that, yes; the government does need to obtain a warrant to gain access to this information. And the unanswered question is what does that warrant have to look like. But, for the time being, the fact that they need a warrant in and of itself I think is very significant. It's a very significant development in the balance between personal privacy and the ability of law enforcement to use every investigative tool available to investigate and prosecute crimes.
Dave Bittner: How could you imagine this in the future having any effect on our day-to-day lives?
Ben Yelin: Oh, man. It's tough to do that when we're talking about Supreme Court cases that are decided on these abstract principles. But it's such a -- it's such a good question, Dave. I mean, I think we can all expect a greater degree of privacy in information we share with third parties online through the use of our smart devices. That includes things like routing information on the websites we go to, on email addresses that we're sending correspondences to. I think it's reasonable, though not by any means definitive, to think that that type of information is going to get a greater degree of constitutional protection going forward because of Chatrie. So I -- I think we have made a major stride in protecting digital privacy, that for information that is conveyed, even if it is conveyed to a third party, that is so deep and broad and kind of central to the use of these smart devices, the Supreme Court has emphasized both in Carpenter and then going much further in this case that we have a degree of privacy rights in that information in a way that we wouldn't have anticipated eight years ago. And I think, no matter what sort of technology you use, I think everybody will eventually feel the implications of that. I think the government is going to have to dot its i's and cross its t's when trying to get that type of information as part of criminal investigations where before they might have just said, Hand it over, Google. Like, I don't need to put together a comprehensive or compelling warrant application. Like a person doesn't have an expectation of privacy in this stuff. They gave it to you, the third-party technology company. Prosecutors and police departments are not going to be able to do that anymore as a result of this decision.
Dave Bittner: Do you think this could change the behavior of the service providers? In other words, could they see this and say, oh, we have to up our game when it comes to protecting this information.
Ben Yelin: I do. You know, I think part of it is going to be a cultural shift that, if the Supreme Court is emphasizing privacy here and in some sense the Supreme Court is reflecting, I think, a lot of real people's concerns about the protection of their private information and how -- how much we're sharing with these companies and the profound impact that that could have on our lives, whether that's through interactions with law enforcement or just through our interactions with these private companies, I think there has to be some acknowledgement on the company's part that privacy is a value. And it's not just a value among the population. It's a value that our highest court takes seriously, especially when we're talking about a certain type of data. So I think with those values and with that new expectation, I think it certainly could motivate the companies to advertise themselves as being more protective of that data, reflecting the spirit of the Chatrie decision. I think that's something that I could certainly foresee.
Dave Bittner: All right. Well, it's -- it's been a lot of fun talking. It's not often we get something this -- this monumental, right, this important.
Ben Yelin: It's not. And I -- you could get me going on another hour about the Gorsuch concurrence and the judgment, which I -- we don't have time for, but I encourage everybody to read. I actually think that to me is -- is probably the most workable theory on these complicated digital privacy cases because it's property-based. We can get to that another time. All --
Dave Bittner: Yeah. Let's save it because I think it's interesting. So maybe next week we can touch on that.
Ben Yelin: Absolutely.
Dave Bittner: All right.
Ben Yelin: I'm glad we got an opportunity to chat about this. And thanks for giving us some content, Supreme Court justices.
Dave Bittner: There you go. All right. We're going to take a quick break. When we come back, Ben's conversation with Brad Carson, former congressman and president of Americans for Responsible Innovation. Stay with us. And we are back. Ben, you recently had the pleasure of speaking with Brad Carson, a former congressman and currently president of Americans for Responsible Innovation, ARI. Your conversation focused on the US-China trade relations and how that could affect AI development. Here's Ben speaking with Brad Carson.
Ben Yelin: So thank you so much for joining us today. It is fortuitous because, as we're recording this, just about an hour or two ago the President of the United States landed in China for this bilateral meeting. So I guess just to start, what do you think is at stake for this trade meeting, especially when it comes to AI, chips, advanced technology?
Brad Carson: I think there's a lot at stake, of course, in the overall relationship between the US and China. But artificial intelligence, in particular export controls, are almost certain to come up. After all, Jensen Huang was apparently a last-second addition to the trip. We know he has a lot of concerns about American export controls. So our hope is that nothing much happens, at least on the export controls. We think the export controls are sound. We want them to be enforced more rigorously. It is a deep fear that there'll be some kind of compromise that allows the top-shelf American ships to be sold to China, and so that's a fear. I think our greatest hope is that, especially now that the White House is seized by what AI can do after Anthropic's release of the Mythos model, that maybe we start having more talks with the Chinese about the future of AI. They're the other competitor in the AI race. It'd be great that we have government-to-government discussions about the future of AI where we can try to jointly shape it to the extent possible that it goes in a good way.
Ben Yelin: Just given your expertise, do you think China and the Chinese government is a trusted partner in a conversation like this, just considering national security concerns? And just can you talk a little bit about how you see that relationship.
Brad Carson: I don't think you should consider them a trusted partner. They're a competitor, sometimes an adversary to us. But that doesn't mean you can't have talks with them. Even when we thought the Soviet Union had global designs on hegemony, we still talked to them about nuclear weapons. So I don't expect them to do us any favors, but I do think that there might be things that are in their self-interest that they might be willing to talk about that we also find in our self-interest, as it has been in discussions with adversaries in the past. So that is my best hope. I don't think you can trust them. I don't think they want our best interests, and so I'm not relying upon them to just do us a favor. There might be an overlapping zone where both of us find it in our self-interest to talk.
Ben Yelin: Getting back to the exports issue, I mean, do you think it's a race to the bottom, like kind of the reverse of or a zero sum outcome if we restrict our AI chip exports to China? The thinking is that we would slow down China and their development of new AI tools, but how do you kind of see that as a factor?
Brad Carson: I think the chip controls are not only the most important, they're probably the only weapon we have in slowing China's ability to develop frontier AI because of the inputs to artificial intelligence data. Well, the Chinese have a lot of data and fewer privacy concerns than the West does. Energy: Energy is going to be a bottleneck in the United States, but it is no barrier to Chinese success. They're building out, you know, gigawatt capacity on a seemingly monthly basis. And the third part are the chips, so-called compute. That's the one area where the United States, in particular, the West, in general, has not only an advantage but really a monopoly. And I said we shouldn't just give those away. That's the only advantage we have. They have brilliant engineers coming out of Ching Wa and Peking. You know, they have all the other attributes of great success. But the chips are holding them back, and they will tell you that they're holding -- being held back by chips. Or the Chinese AI labs will tell you that, hey, the chips are a barrier to us being more successful. So the idea that we would then give them the chips is like, you know, as Khrushchev would have said, you know, selling them the rope to hang us, it makes no sense at all.
Ben Yelin: Do you think, just from a diplomatic perspective, it seems like that, you know, there are rumors of Chinese firms routing chips through third countries or through some of our other adversaries. I mean, do you think a lot of the work that needs to be done is diplomatic, just making sure that we maintain that competitive advantage in exporting chips?
Brad Carson: Yeah. I think three things need to be done. The first is, yes, more diplomacy. So we know that chips, for example, are being routed through Singapore, Malaysia, other countries like this so greater diplomacy with them to crack down on it. The second thing is I think it's widely believed, in fact, mostly known that Nvidia turns a blind eye to that kind of smuggling. Jensen Huang doesn't support the export controls and, therefore, is kind of blithe about the smuggling. And so Jensen Huang should step up. Nvidia is an American national champion, right? They should prohibit this. And the third thing is Congress now is making moves to stop this because there's a number of bills, bipartisan bills in the House and the Senate to try to have some kind of location verification so that the chips themselves can be controlled and you know where they are at all times. So I think that is what the future is going to bring, so all three of those things are necessary to stop the illicit traffic of top chips to China.
Ben Yelin: When you think about the major players in Silicon Valley, kind of the David Sachs types, the innovators, how do you convince them that these types of export controls are in the best interests of the industry long-term?
Brad Carson: So, if I divide the question into two, on export controls, I meet very, very few people, unless they're on the payroll of Nvidia, who don't support export controls. I mean, this is true for the far right, the far left, the moderate. You know, everyone supports export controls except for a very, very few people, again, around Nvidia, maybe David Sachs. I honestly have not met a single member of Congress. My job is to talk to them every day. I've talked to hundreds of members of Congress who don't support the export controls and who tell me, like, you know, we should relax those. None do. Now, the question of guard rails domestically is a different one. And I think the case there, people would say, like, hey. We need to innovate. Maybe we need to out-race China. You know, some people are deflationary of what AI can do, that we don't -- it's not that dangerous. It's not that problematic. So why would you have to regulate it in some way? There's actually a surprisingly large number of Silicon Valley people who are like, Yeah. I'm fine to merge with the machine. I'm fine if the machines are a successor species to us. That seems absurd to normal people. But, in Silicon Valley, there's actually like a type that believes that. And so each of them require, like, a different dialog about it, you know. If you want to merge with the machine, I don't have a lot to say to you. I'm like, okay. Right. I get -- I get where you're coming from. I'm just going to oppose that project. I think for people who are more, well, in the mainstream, I would say that guard rails and innovation are going to go hand in hand because right now the kind of free for all we have in AI is leading to a massive backlash. People with pitchforks are coming for data centers. AI is deeply unpopular; 80, 85% of Americans hate AI. I don't know if you saw recently there was a commencement speech at the University of Central Florida where the commencement speaker said something along the lines of AI is the next industrial revolution, and the students booed her.
Ben Yelin: I did see that. Yeah.
Brad Carson: And so we're in a place where, like, if we don't have guardrails, the American people are going to shut this whole project down. And so it's like air -- it's like airline safety or even traffic safety. The guardrails, the rules, the yellow lines down the middle are actually what make high-speed traffic possible. And so I think the guardrails innovation are not orthogonal to each other. They're, in fact, kind of in parallel.
Ben Yelin: This might be a bit of a niche interest, but I'm very interested in this question of federal preemption. I know the administration tried to include federal preemption in the Big Beautiful Bill, and they've made another effort with an executive order to preempt state laws on artificial intelligence. I just wanted to know your perspective on that.
Brad Carson: Now, we've led the group I run, Americans for Responsible Innovation, the opposition to a blanket preemption of state laws for a couple of reasons. We think there should be a comprehensive federal regulatory scheme about AI. And, if that is in place, then one can honestly talk about, you know, what state laws might conflict with it. And you could imagine some kind of preemption. But what the other side is really argued for is like, we don't want anything at the federal level; and we will fight you if you try to try to propose that. And we're going to try to also preempt all state laws. So you have this regulatory black hole. It's really the same strategy that the cigarette companies did for a very long time at the 1960s and 1970s. And so, no; you should not have a blanket preemption absent a corresponding federal scheme that regulates AI. And so my vision for how this should go -- and I think we're going to get there -- is the federal government has certain things that are important to it, national security ways of using AI, for example, cyber weapons of mass destruction. The states have legitimate concerns about how AI is being used in, say, employment benefits determination, criminal justice; and the state should regulate that. They both have spheres where they're working, and that's the way most industries are regulated in this country. You think of, like, vehicles again. Car design is regulated by the federal government. The states can't tell Detroit how to, like, design a car. But, in every state, you have different rules about how the roads work and who can drive; and emission standards can differ in state to state. So that's how the world should go. But right now the fight has been in people just saying we want no regulation at all. And the states, to their credit, aren't waiting. They're more agile. So they're actually trying to regulate AI. And that's a good thing, absent any kind of federal activity.
Ben Yelin: Yeah. And it's been interesting to see it's not just the blue states, the normal ones you'd expect to regulate like California and Colorado. It's a lot of red states. I mean, Tennessee was at the forefront with the Elvis Act, among other things. And I know you, as a former Oklahoman and congressman, can appreciate that, the bipartisan nature of this.
Brad Carson: It's very true. I think, if anything, it's the conservatives. You know, they're fiercer about AI than the progressives are. Steve Bannon is like a hobby horse for him. You know, he hates AI; and he's on about every day about how dangerous AI is to us. And it really is a bipartisan effort. The faith groups are strongly suspect of AI. So, yes; Bernie Sanders is against it, but so is Bannon and Marjorie Taylor Greene. And so it's a very bipartisan group. And, again, it represents about 80% of America, actually, who are like, yeah. It has some good aspects to it; but I'm kind of worried that we're going in the wrong direction with it. And I think that impulse is right.
Ben Yelin: Right. And I think, you know, you kind of mentioned the full-on Silicon Valley view of I'm fine handing over control to my life and my destiny to AI versus -- and they'll say, like, we can supplement it with a universal basic income, and everybody would be fine, versus the reality that you mentioned that people are very upset about data centers using energy and water. People are very upset about automation, that sort of thing.
Brad Carson: I think the data centers are in many ways a proxy for AI. It's like it's how the -- how a normal person sees AI coming to their community and how they can register their protest against it. And so -- and that's not going away. Again, both conservative and Democratic states, I come -- and progressive states. I come from Oklahoma. Still live there. Both Tulsa and Oklahoma City have moratoria on data center development. And these are not left-wing cities. They're not San Francisco or New York, but they're very conservative cities. Evangelical Christians dominate, but there's just a lot of skepticism. And that's why the guardrails are essential because the truth is America has lost faith in its tech sector. And the tech sector has to work to regain it, or else the project is going to fail. And the project's failure would be actually a terrible outcome for us all. Right? We don't want it to fail. But it must be done in the socially responsible way, or our immune responses will just fight it.
Ben Yelin: Right. I think that's very well said. I want to just, given your wealth of experience, to get into a couple of sector-specific things. So talking first about the military and defense, just given your work at the Pentagon, what's your kind of level of fear about AI and the use of military readiness and offensive and defensive operations. Are you concerned about the application of AI in military affairs?
Brad Carson: I am concerned. Again, it has some great use cases in the military. So imagine you're getting terabytes of intelligence, you know, imagery coming in from all the drones, all the signals intelligence from all the collection devices, you know, all the human intelligence reports, you know, all this kind of intelligence coming in. No human can assimilate it all into a common operating picture. AI can do that. I think that's a great use case for it. And that was actually the original use case of Project Maven. You know, yes; I think that's a great use case. I am much more worried about the idea that people are being identified as possible targets on the basis of AI because, you know, as I often talk to people, what's happening in a place like -- and we've seen this in Gaza, you know, where Israel has used a lot of AI to identify targets, right. They're assimilating incredible amounts of intelligence, also facial surveillance data, you know, human intel, all these kind of things. And every one of us has a score. Like, Ben, you have a score, like on a gradient that you know there's an 80% that you're a Hamas terrorist. And it's based on this kind of Palantir kind of technology that's using this. And so the issue of what's that even mean, you know, to be like an 80% chance of a terrorist. This is the big change. Historically, it's been a categorical determination. It's like, Ben, you're a combatant and, therefore, I can strike you; or you're a non-combatant, and I can't. Today it's an 80% chance you're a combatant or a 70% chance. And who picks that number, by the way? And is that number, what's it even mean when it's coming from a neural network? These are things, you know, we're now all on a heat map. And, you know, if you're in the wrong color, you could be hit by a drone. And I think that's, like, actually a really dangerous change. And we know from social science that people defer to computers a lot. They don't question it. It's happening in, like, incredible rapidity. The targets are being generated rapidly through this system. I think these are all things that should be quite worrying to us all.
Ben Yelin: And then, on cybersecurity, I mean, this is something that worries me significantly. Do you support pairing expert controls with stronger cybersecurity requirements for our chip manufacturers and distributors? And can you just talk a little bit about the cyber risks, just given the prevalence of cyber warfare that we're seeing now, including during this conflict with Iran.
Brad Carson: Yeah. There's two issues that I think are invoked by your question. One is we know Mythos is out there, and it can find lots of zero day defects and has incredible cybercapabilities. But you suggest something that's less talked about but, to me, incredibly worrisome, which is, if you think OpenAI, Anthropic, Google, maybe X AI, maybe Meta, right, are creating world historical products that have incredible national security application, then you have to assume people are trying to steal them, you know? I mean, Xi Jinping would be derelict in his duty to not try to steal them in some way. And we know there's lots of people at the labs, right, who have ties back to these other countries, as well, people who work there. Many of our best engineers, after all, right, are people who come from, say, China, you know. And that's not to say that it's loyal in some way. It's just to say the industry is very internationalized. And so do the labs today have kind of national security grade cybersecurity to ensure that people can't exfiltrate the weights? They do not have those kinds of things, you know. So these are the real concerns you have is like China doesn't have to develop Mythos. They can steal Mythos. And we have to assume they're trying to do that. Again, China would be derelict not to be trying to steal Mythos in the same way if China were to develop something like that, if you're the US government, you should be trying to steal it, you know, or at least maybe disable it, even. But probably just steal it. And do we have the protections at the labs? We do not have the protections at the labs, you know, the cybersecurity, the physical security that it takes to protect them. That's a grave concern. And, yes; you know, we spend a lot of time advocating to the government, saying, like, we must ramp up the cybersecurity and physical security at these labs because they're -- they're national assets. It's like -- you know, it's like Lockheed doing stealth or something. Like, yeah. You have to keep this, you know, where they can't steal it from you.
Ben Yelin: Right, right. Turning to be a little more forward-looking, 5, 10 years from now, how would you define us winning the AI race, whether it's just this kind of US versus China geopolitical battle or other adversaries of ours get involved. Is it technological dominance? Is it deterrence? How would you define us -- I know it's -- you can't really bifurcate winning and losing in this context, but just how would you look to define it going forward in the next 5 or 10 years?
Brad Carson: To be honest, I would reject the framing of a race. I don't like that at all. I don't think a race is a way to think about it because what's the end line? You know, what do I get if I win the gold medal? You know, what lane am I in here? You know, I don't know what winning looks like in this kind of context, you know. It's -- it's an arms race of sorts, I suppose. But we know from history that arms races, it's usually a negative pejorative term people use. It's never good to be in an arms race. You know, they're expensive, destabilizing; and you should try to get yourself out of it to some extent. And it often leads, if you have this framing, to kind of being bellicose and overinvestment and national security concerns that, like, take over kind of more rational thinking. I don't think there -- the race framing is the right one. We're in a competition, but it's a competition that has a number of fronts. Yes; we need to develop frontier AI. We have a lead in that. We also need to actually put AI into our economy through, like, robots. China is going to have a lead on that, you know. We have to, like, ensure that we have a social safety net that can absorb this technological shock. We're way behind there too. China is also behind and concerned about that. So it's like a number of fronts. I think it's more like about seeing, like, the industrial revolution. You know, was England in a race with France or Germany or -- or even China back then? Not really. You know, it's about developing tools of national competitiveness. It's about sharing them to some extent, you know, to accept they can have healthcare innovation and medical breakthroughs, material science breakthroughs. I think the arms race framing is a militarization of AI that I think in general should be resist. And it's not -- to me, it's a bit incoherent to think of it in that way, and it's also a very dangerous way to think about it. But it is, sadly, the way most people in DC now think about it; but I don't love it myself.
Ben Yelin: I'm just wondering, along with that, because I agree with you, and I think, you know, it certainly brings up a lot of interesting thoughts. One thing that struck me is that in no -- no other industry do you see the titans of the industry talking about how dangerous their product is and how it could be like a civilization-defining event when general artificial intelligence is developed. How do you work within that context where there is that level of fear, not just among the general public but being expressed even by the titans of this industry? And how does that impact the balance you're trying to strike between leveraging the good uses of this technology while also protecting our interests and our security?
Brad Carson: I think a couple of things are notable in what you said. You're right. It's the titans who tell us that. And I think they're being sincere, by the way. You know, there is some discussion that maybe they're talking their book and trying to boost their IPO. No. They're actually believers. And, when you go out to San Francisco and you talk to the 25- to 35-year-old engineers, the people who are actually making it day to day, their fear is palpable and real. And they're not faking it. They're not just, like, putting on a front about this. They're like, no. If you knew what we knew, you would be like really scared that this is developing so rapidly with such capability. It's so transformative, and we're not prepared for it. So I think that's a key part of it. And I think the second issue is we need to take those warnings seriously and -- and realize that we have to respond politically because the big issue the labs in America have now, to me, it's a collective action problem. If you're in a -- there's not a race with China, but there is a race among the labs between Anthropic and OpenAI, Google especially. And, you know, I've heard Demis Hassabis, who runs Google DeepMind. I've heard Dario Amodei say both in Davos, like, you know what? We would be favorable for a pause, for a slowdown of this. But we can't do it by ourselves because, if I do it, you know, OpenAI is going to race ahead. If OpenAI stopped, Anthropic would race ahead. So there's a great role for, like, leadership to step in and say this is a dual-use technology, has many positive aspects, negative aspects. Let's solve the collective action problem where we're not all just, like, fearful of being defeated in the market. And the government could actually step in. It doesn't even have to have a law, like a norm setting. You know, President Trump could have a meeting in the Oval Office with these lab leaders and say, Let's all, like, bless this approach; and they would probably do it. So I think that's what we do is we have to take these people seriously. They're not faking it. They honestly believe they're developing a world-changing technology that has a lot of dangers, and they're actually making real progress to achieving that goal.
Ben Yelin: Right. As dubious as that is.
Brad Carson: Yeah. I mean, I often say to folks, like, they say they're doing it. They're making meaningful progress. They're devoting unprecedented amounts of capital to this project, and the smartest people in the world have all congregated in a 6 x 6 square mile area in San Francisco to do this. And I said, like, yeah. It's like a potent brew. And we should be, like, really, really scared about it and think, like, it's for real.
Ben Yelin: I think that's exactly right. Very well said. Brad Carson, former congressman, the current president of Americans for Responsible Innovation, thank you so much for joining us today.
Brad Carson: My honor to be with you.
Dave Bittner: Yeah. Interesting stuff, Ben. I mean, this -- it seems to me like, even in just the past couple of weeks we've seen stories about these frontier AI models and the possibilities of China catching up and what that could mean, and it really seems like China is really the only other big player here.
Ben Yelin: Yeah. I thought it was a really interesting interview. He's doing really interesting work because I think he's trying to balance, like, the AI optimist view with reasonable concerns about regulation. And he's trying to kind of work in a -- in a -- in a middle space between the two extremes of, like, we need to preempt all state laws related to AI versus AI is going to kill us all.
Dave Bittner: Right.
Ben Yelin: Like, let's ban data centers.
Dave Bittner: Right.
Ben Yelin: And so I admire him for doing that. I thought it was a really interesting conversation.
Dave Bittner: Yeah, yeah. Well, again, our thanks to Brad Carson from Americans for Responsible Innovation for joining us. We do appreciate him taking the time. And that is Caveat brought to you by N2K CyberWire. We'd love to know what you think of this podcast. Your feedback ensures we deliver the insights that keep you a step ahead in the rapidly changing world of cybersecurity. If you like our show, please share a rating and review in your favorite podcast app. Please also fill out the survey in the show notes, or send an email to caveat@n2k.com. This episode is produced by Liz Stokes. Our executive producer is Jennifer Eiben. The show is mixed by Tré Hester. Peter Kilpe is our publisher. I'm Dave Bittner.
Ben Yelin: And I'm Ben Yelin.
Dave Bittner: Thanks for listening.
