New Jersey’s former attorney general on Ring cameras and facial recognition

In this week’s Vergecast, former New Jersey attorney general Anne Milgram stopped by the studio to talk with Verge editor in chief Nilay Patel and me, senior reporter Colin Lecher. As Nilay notes, Milgram, who also co-hosts the podcast Stay Tuned with Preet Bharara, is “the first cop we’ve ever had on the show,” and gave some thoughtful responses to questions about surveillance, predictive policing, and more. “We all, I think, have the right reaction, which is we don’t want to use data that’s biased or we don’t want to have problems,” Milgram says. “And yet in our personal lives, we give access to a huge amount of information and a lot of it is not public.” The rise of home security systems like Amazon’s Ring camera have raised serious questions about privacy, and Milgram weighed in on the issue. Below is an excerpt for that conversation, lightly edited for length and clarity. Learn more about your ad choices. Visit podcastchoices.com/adchoices

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:01Speaker 00

  Company spending is complex. Managing it should be simple. This is Amex Corporate, a single platform connecting cards, expenses, and payments to streamline the way you work. Flexible controls for when priorities shift, support around the clock, all backed by American Express, so you can stay focused on what matters most. Built to simplify, designed to scale. This is Amex Corporate. Get started at go.amex slash Amex Corporate. Terms apply.

• 0:35Speaker 01

  Your heart can tell you a lot about your health. Apple Watch Series 12 measures your heart rate every five seconds with the most accurate heart rate sensing in wearable. So your vitals app, now with heart rate variability, can tell you when something is off. And your readiness score can let you know when to rest and when to push. Hear the story in every heartbeat with Apple Watch Series 12. The features described are for wellness purposes only and not for medical use. iPhone 11 or later required. Based on Apple conducted study of heart rate accuracy, August 2026. Visit apple.com slash Apple Watch Series 12.

• 1:05Nilay Patel?

  Hey, everyone. It's Sam from The Vergecast. On this week's interview show, Colin Lecher and I were joined by Anne Milgram, the former Attorney General of New Jersey and a co-host of Stay Tuned with Preet Bharara. Anne spent a lot of time figuring out how technology and law enforcement work together. So we talked about things like private surveillance networks, like the Ring cameras, and how they feed into things like predictive policing, algorithmic bias, the use of data to prevent crime instead of just punishing criminals. This is like the first cop we've ever had on the show, but it's a good one. Anne is a really interesting and thoughtful person on these issues. She really made me rethink some assumptions I had. I thoroughly enjoyed this conversation. Check it out.

• 1:45Nilay Patel?

  Colin Lecher, how are you doing? I'm doing great. How are you? I'm good. Thank you for asking. No one ever asks. We're here with Anne Milgram. You're a professor at New York University.

• 1:54Speaker 03

  Yes, School of Law. Yep.

• 1:55Nilay Patel?

  You used to be the Attorney General of New Jersey. You worked for the DOJ. You worked for the Manhattan DA. And now you host the Cafe Insider podcast with Preet Bharara.

• 2:03Speaker 03

  Yes.

• 2:03Nilay Patel?

  That's a lot of things.

• 2:04Speaker 03

  It is. I'm tired.

• 2:06Nilay Patel?

  My note here says you focus on issues of law, politics, and justice. You also, we were just talking, you focus a lot on integrating data across those things.

• 2:13Speaker 03

  Yes.

• 2:13Nilay Patel?

  Tell me what that means exactly.

• 2:15Speaker 03

  So I started, if you want to go back to sort of the data tech piece, which is I was trained as a lawyer, became a prosecutor, was at the Manhattan DA's office, then at Main Justice, and then I become AG of the state of New Jersey, which is a unique state, as you know, in many ways, including that the AG is the chief law enforcement officer. So, one of the prior AGs had taken over the Camden, New Jersey Police Department. And so on day one, I was 36 years old, and I was in charge of the Camden Police Department, and it was the most dangerous city in America. And so I went down, and, you know, the best way I could describe it is, like, you know when you get those eye tests and they dilate your pupils and everything is blurry for like four hours? It's like, I, literally, everything was blurry to me in Camden.

• 2:58Speaker 03

  I just, I couldn't understand why people were dying, why, you know, I went down, there were no police officers on the street the first day I went down. I went to a comstat meeting, which is where departments use data and analytics to try to understand where crime's happening, where the police officer's being deployed, like, essentially, how are we responding and preventing crime? And the short answer is, we weren't. And we really didn't know what we were doing, we didn't know if it worked, and we didn't know if we could do it better. And so, that really started, and incidentally, I was sworn in as AG, June 29th of 2007, which is also the day that the iPhone came out. Oh wow. And so, it was sort of like, this world, the world was changing rapidly around me, and I'd been to the NYPD comstat, once, when I was an assistant DA in Manhattan, and just

• 3:42Speaker 03

  sat in this room and watched how data was helping them understand robbery patterns in a way that they'd never been able to do so before. Before, it was just isolated, this officer made an arrest for robbery, there was a 911 call for robbery. But now, they were mapping the city and understanding how things were happening, and they weren't happening in isolation, so data really was letting them see the truth of what was happening. And so, that leads me to this, this experience in Camden, where we spend a year, essentially, pulling data, understanding where the officers are being deployed. We had a ton of 911 calls when I got there, and just understanding, in a city of less than 80,000 people, why do we have over 11,000 or 12,000 911 calls? So just trying to really drill down, and then asking this fundamental question, which

• 4:26Speaker 03

  is, communities with violence or crime, all, you know, which is essentially all communities... It's not random and it's not evenly distributed. There are a handful of people who commit the majority of violent crime and a handful of groups often, it could be a gang or a drug trafficking organization. And so just thinking about how do you gain insight into those organizations and those people and really focus your time and effort, not equally on, you know, the person drinking the 40 on the street in the middle of summer and the person who may shoot the next gun, but really thinking about how do you stop the next shooting? And so it took us about a year, a year and a half to get on top of what was happening, to completely redeploy the police department, to shut down open air drug markets, to focus

• 5:10Speaker 03

  on the hotspots, the area where the crime was really occurring. And in one year, we dropped violent crime by over 40%. We dropped homicides by 46%. Camden today, this is, you know, 12 years later, it's not perfect, but it has done extraordinary in reducing crime. It's a million times safer than it was. And so data really, to me, was the sort of turnkey in understanding what was happening and understanding how we could start to police in a way that would make the city safe. And so that, that sort of is my introduction to data and technology. And I tell people this all the time with police departments, I meet with a lot of police departments, there was really no tech at the beginning. I had an amazing guy, Joe Cordero, who'd run the East Orange Police Department, and he

• 5:54Speaker 03

  was obsessed by, by tech and data. And he literally ran ComStat, which is basic statistics, right? About, you know, let's look at the last five years, see when the crimes are occurring and where they're occurring. He literally just did it by hand for the first year. Like, we didn't buy a computer program, we just, like, he just went in.

• 6:09Nilay Patel?

  Yeah, Google Sheets is free.

• 6:11Speaker 03

  Well, now.

• 6:12Nilay Patel?

  I'm just saying.

• 6:14Speaker 03

  I don't think, I mean, I don't know, I probably had like, I think I had, I'm gonna admit this, and I shouldn't, I shouldn't admit it, I'm sure I had MSN back then.

• 6:21Nilay Patel?

  Wow.

• 6:22Speaker 03

  I know.

• 6:22Nilay Patel?

  That's, I mean, that's great. Like, those email addresses are like a sign of, like, oh, genus. Like, I still have my ale.com email address.

• 6:29Speaker 03

  So does my mom.

• 6:30Nilay Patel?

  Yeah, that's great. But there's like, there's layers. And I, you have a TED Talk about this stuff, and I was reading some stuff you wrote for The Atlantic. There's obviously where to deploy the police.

• 6:40Speaker 03

  Yep.

• 6:41Nilay Patel?

  Where they should go, there's who they should arrest, there's when they should arrest them, who you should prosecute, what you should charge them with, when to take the, like, there's layers and layers of sort of the law and order stack, right?

• 6:52Speaker 03

  Yes.

• 6:52Nilay Patel?

  The criminal justice system, where is the data most effective, do you think? When you're rolling it out this way?

• 6:58Speaker 03

  That's a great question. So first of all, don't tell pre, but my best job is that I work for Law and Order. So we'll come back to that. As I always say, I try to make the shit, like I work with the writers on the criminal prosecution side to make it as legal as possible. Where's the data most effective? So if we, if we really go 30,000 feet and talk about the United States of America, we have terrible data in criminal justice and it is a disservice to everyone. It's a disservice to people and communities, to the police, to victims of crime and to people who commit crime and are being held accountable. We have just so little idea of what we're doing, whether or not it works, and really, these fundamental questions of who comes back after people have been arrested? What are the things we can do to reduce crime the most in the future?

• 7:43Speaker 03

  And to me, how do we prevent future crime? You know, to me, it's all about public safety is preventing future crime. Right now, so the sort of short answer is right now in America, I would like everyone to collect basic data just to understand what's happening, where the, like just that first cut of data and, you know, that Gartner sort of data chart where at the end you get to, you know, sort of, you're like, it's actually, data is changing the way you, you live and preventing crime and stuff. But the first is just the basic understanding of how your system works. Who's coming into your system? What are they coming in for? What are your officers? I mean, I, I have this conversation with both police chiefs and DAs all the time. What do you spend your time on?

• 8:24Speaker 03

  Like what is, you know, if we, if we did this graph of 100% of, graph of 100% of your time, where are you spending it? And it's like crickets, right? Because most people, they don't have that data and just, so people understand, police departments collect data for their internal management. DA's offices collect data. Same thing with courts. Everybody's using it to drive their internal systems, not to understand the criminal justice system as a whole or really. How they're spending their time and resources and how they can be more effective. And so I think, you know, my first, and the big cities are different. The big cities generally do a much better job of this, but you know, there's 10 massive police departments in the United States, maybe 20, and there are thousands of police departments

• 9:08Speaker 03

  across America, 20,000 or something. And so if we really think about it, we should be asking this question of like, how do we just gain basic insight into how we're running the day to day of our operations?

• 9:19Nilay Patel?

  So let me zoom out even farther from data and tech. You've said effectiveness a couple of times now. I think we're in America at a moment of reconsidering what effectiveness means for a police department. How would you define that?

• 9:33Speaker 03

  That's a great point. So, yeah, and I think, I think this is the right conversation to be having at this moment in time when our rates of incarceration in the United States are just, obviously, they're so much higher than any place else in the world. And the stat, we have the highest rate of incarceration, meaning per 1,000 people, we incarcerate the most people around the world. And the stat people uses, we have 5% of the world's population, we have 25% of the world's prisoners. My issue with that is that when I was at the Arnold Foundation, I ran a criminal justice initiative for the Arnold's. We did a study that looked at what happens when people are detained prior to trial, just this, this critical point in time before people have actually been, you know, adjudicated on, on the crime that they're arrested for. Are you guilty or innocent?

• 10:17Speaker 03

  So people are still presumed innocent. And we found this, this amazing thing, which is that people who were low risk and moderate risk, meaning not that likely or moderately likely to commit a new crime, they ended up, if we incarcerated them, committing more crime, both pre-trial and in the long term. For people who are high risk, those are people who are likely to commit new crimes. And by the way, just for folks to sort of be able to understand this, that's a very small, it's a really important part of the population, really important for public safety, but probably about 10% of the entire population that goes through the criminal justice system is high risk. 50% say is, is low risk and maybe 40% moderate, give or take, depending on the jurisdiction.

• 10:58Speaker 03

  But so you're talking about essentially 90% of the people do worse when they're incarcerated. And they do, that means that there's more crime, both prior to their trial and after their trial. So if we really care about public safety, we have to be having this conversation of how do we handle? When people break our laws, like what, how do we hold people accountable? But also not necessarily hold them accountable through jails and prisons, which we know the outcome is generally more crime, which is the last thing any of us want. And so how police officers internalize that, it's, this is a complicated transition. And it's complicated because a lot of people go to police departments because they want to be crime fighters. And I can tell you, you know, all the officers that, that worked with us, like, they always

• 11:40Speaker 03

  wanted to wear like, you know, the raid jacket that says like FBI or NYPD on the back and they want to take down the door. But a lot of policing is about just being in your community and just walking the streets. And then there's also really sophisticated policing, which is about understanding who is planning, you know, who's planning an attack on a rival gag? Who is, you know, there's a group of, of people doing break in robber, break in burglaries or break in robberies, like who, who are those people and how do we get ahead of that? You know, I'd like to tell the story, my grandfather was chief police of a tiny, tiny town in New Jersey, one square mile, famous in the Guinness Book of World Records at one point in time for having the most bars per square mile as one.

• 12:23Speaker 03

  But his job, this was before 911 calls, he used to just walk the streets and there, there are a lot of bars in South Amboy. I hope the mayor doesn't call me screaming, just a fact. I'm from Wisconsin. It's just truth, right? And so, you know, my grandfather would walk when the bars were closing and just make sure everybody got home safe. Right? And it was like, the police was a real part of the community versus being sort of at odds with the community. I also think that we have to be very honest about what the police can and can't do. And so there's, I spent a lot of time, we just built a tool for police officers to use that will screen people on the street for mental health, substance abuse and homelessness. With again, this idea being that there are a lot of people in the system who have very,

• 13:05Speaker 03

  very high rates of those three things and the system doesn't treat them. And so if we want less crime, we have to figure out, how do we get those folks treatment? And so we built this tool, we're beta screening, we're beta testing now with Indianapolis in Indiana and McLean County in Illinois. But the whole idea is like, how do we figure out, how do we solve some of those problems? Cause if we don't solve them, they're going to come back again and again to police. And the police response really is. We arrest you, you know, it's a crime and punishment system and we should be honest about that's how we built our system. You know, someone engages in something that's criminal, the police arrest you, the prosecutors convict you, and you get incarcerated. And that model just doesn't prevent crime for a lot of folks. It doesn't, it doesn't make us as safe as we, as we can be. And again, there's some people who need it, but we haven't

• 13:50Speaker 03

  done a good job of figuring out. It's a funnel now and we put everyone through the same funnel. And I think the real obligation of the police going into, you know, the next decade is to figure out how do they keep people out of that funnel effectively? And how do they figure out who the right people are to put in?

• 14:05Speaker 04

  I think one of the concerns that people have with these sorts of systems, especially in the past few years, is, you know, we have this data, but maybe the data relies on police already looking at, say, like, poorer neighborhoods or communities of color, and then we're re-putting that data into the system, and then we're just exacerbating the same problems that we've had for decades. And how do you, when you're building a tool or trying to collect that data, how do you make sure that that doesn't happen?

• 14:33Speaker 03

  Yeah, so that's a, that's a great question. And so, again, I think if we go back to how we think about data, the first thing is just understanding what, like, some of that argument has stopped police departments from collecting data overall and being ruthless and understanding what they do and what they don't do. And the way I would put it is, and I think this still exists in a lot of America today, it's very anecdotal. Policing is very anecdotal and is sort of like, you know, I think I know where the crime is, or I sit on one particular neighborhood because I think that there's a lot of crime here. And you end up, if a police officer's in a neighborhood, you end up seeing more arrests in that neighborhood. And so, it's a little bit of a, it's not a little bit, it is a vicious cycle. So, the question is, how do you sort of get over that anecdote? And how

• 15:17Speaker 03

  do you think about, frankly, holding people accountable in a way that's fair and gets to the point you want? So, the first cut of data, I think, doesn't implicate these concerns as explicitly as the, the use of that data, right? And sort of using the data to make decisions. So, the first cut is just, what are we actually doing? Who are we arresting? Are we disproportionately arresting people of color? I work with a city, and we just ran a ton of their data, and... We were looking at, primarily at, mental health and found that people who suffer from mental illness, we matched them by, you know, age, gender, crimes, they were saying significantly longer in jail. And there's no reason to believe that the prosecutors, or the police, or the judges

• 16:00Speaker 03

  were really aware of the mental illness or making decisions based on it. It's just, the outcomes are dramatically, twice as bad for folks in those buckets. So that's the kind of thing that's important to understand. We also found, when we pulled the data, that people of color were, on average, spending 8.5 days longer when they were incarcerated. And again, we matched on everything, right? So, there's huge bias in the criminal justice system. And I think, one of the things we all have to be really brutally honest about, is that there is racial bias in the criminal justice system. And we have to understand that that is the world that exists right now. So people are detained longer, and again, I'm not, I'm not saying it's knowing, even, or intentional, I mean, we've seen it with mental health, we've seen it with people of color.

• 16:44Speaker 03

  So, I think it's really important to start at the point of understanding that we have a biased system, and it's not just the policing system, right? So that's, that's step one. And by the way, I acknowledge, I'm a part of that system, and so I, you know, I take responsibility, but I also think it's just really important to be blatantly honest about that. The second question is, can we improve that system? And how can we improve it? And can data be effective and helpful? And people in minority communities are often overpoliced, they're arrested more, so the data has this implicit, and actually, it's explicit bias, right? The data has bias in it. The question then becomes, what do you do with that? First of all, how do you use the data? Do you use the data to detain people? Do you use the data to release people? That's a real question.

• 17:28Speaker 03

  What data are we comfortable using? You know, I've worked on policing artificial intelligence projects where we just didn't use misdemeanors, where we didn't use any drug arrests, right? Because those tend to have the highest rates of bias. And so, again, the data turned out to be really valuable in telling us what was happening, where was it happening? How would you think about policing a community? But, we were able to take out a lot of things that we think historically caused the most problems. What I would argue should be happening, because I think this is a really big problem, but I also, I would argue very strongly against it meaning that we shouldn't use data. Because if the choice is between an anecdotal decision that we know has bias already baked

• 18:10Speaker 03

  into the foundation of it, and a data-driven, you know, piece of information that again, doesn't negate someone's personal instinct or intuition, but is used in concert with that, the question then becomes like, what data are we comfortable with using? What are the rules of the road? What data shouldn't we use? And no one is having that conversation. And that to me is like, we shouldn't walk away from all criminal justice data. The question is, how can we use it? What can we use? And truthfully, what shouldn't we use? And what should we just basically say, it's never okay to use this because there's too much bias?

• 18:41Nilay Patel?

  Yeah, I think you're the first cop or cop-adjacent guest we've had on the show. I don't know, there's been a lot of guests, but I'm fairly confident in that, in making that call. And usually I get like startup CEOs or tech CEOs. And like the CEO of a keyboard company, when he's like, we read the data and we decided that the keyboard should be green, like extremely low stakes.

• 19:02Speaker 03

  Right.

• 19:03Nilay Patel?

  Like zero stakes decision for Amazon to be like, the box will be black instead of white, or whatever. You are making extremely high stakes decisions. It seems like that process should be a little bit slower, but also there should be a clear understanding of the stakes. Obviously, that's your project, but do you think other people understand? Because I see this over and over again. I see it in our organization, I see it in the organizations we cover. You give people a little bit of data, and it feels objective, it feels real, you can't really argue with it sometimes. It's like snap decisions get made all the time based on one graph at a presentation. How do you both inject the data to increase rationality, increase effectiveness, however we're going to define it, but then prevent that sort of snap decision that creates the bad feedback loop column was talking about?

• 19:46Speaker 03

  Yeah, I think there's a couple of things. And I think, first of all, this really is a problem that society as a whole has to solve in terms of having an honest conversation about what data we use, what data we don't use, and how do we address the existing bias and try to basically create systems that don't, you know, the bias is baked in right now, right? And so it's a very, it's a very tough thing to do. What I think about is You know, police departments are pretty hierarchical, right? And they sort of, you're an officer, then you become a sergeant, then you become a lieutenant. Like, it's paramilitary in some ways. And they traditionally have been about in the morning, or whenever you go on shift, we do a roll call meeting where we say, here are the threats. Generally, we've had, say hypothetically, we've had, you know, a rash of robberies, and we want you to be looking in these three neighborhoods.

• 20:31Speaker 03

  And then we send officers out for 12 hours. And what do they do? They run after 911 calls, right? And maybe the detectives do some work around solving a homicide. Clearance rates in America are actually pretty low for murders. Not in New York. New York is pretty high, honestly, but most of the United States is pretty low. So how do you then take the way that those operations are run and sort of almost educate leaders as they're coming up into, how do you think about policing using some data? So we, as part of Camden, we hired some analysts who could basically say, okay, here is the pattern of robberies we're seeing. They did CCTV, and they rotate now through this, they call it the Real-Time Intelligence Center.

• 21:16Speaker 03

  They rotate junior officers in so they could actually watch on the screen. They see, okay, there's a report of a robbery in this area. They turn the cameras to see what's happening. And there are some areas that aren't policed, but they have cameras. And then they deploy officers consistent with that. So they're sort of ingratiating the officers who are coming up into understanding there is value to data and technology. It's not the be-all, end-all. There's no substitute for the decision that people make. And, you know, if left to it, the officer on the street, you have to trust their judgment. We just have to figure out, can we inform their judgment better? But I think, right now, the way it's working in policing is that there are private companies who come in and sell all this tech to companies. And I'll give you a great example. In Camden...

• 21:59Speaker 03

  When we were the most dangerous city in America, we went out to a bunch of companies and said, okay, we think we need CCTV, we think we need to build a comstat model. And the bids were like $3 million, $4 million. And at the time, and this is still true in a lot of America, the companies would have owned our data. So we wouldn't have even had access to our data. Instead, we had this amazing guy, Joe Cordero, who came in and he just designed it and built it. And we built it, I don't know, it was under $1 million ultimately. And we captured, we were able to keep our own data, keep an understanding of sort of what was happening. But most police departments go the former route because they don't have somebody who can sort of straddle those two worlds. And then they end up with these systems they often don't know how to use, they don't have access to their data.

• 22:44Speaker 03

  And it's actually not teaching them this sort of conversation we're having of like, how do you use data to help you understand? How do you make communities safer? And how do you make, how do you build more justice? Right? And, you know, my view is ultimately we can have less incarceration and less crime, but we have to be really smart about how we do it. And I've not seen a lot of models that are successful yet.

• 23:05Nilay Patel?

  The nightmare IT vendor overselling you crap is like, already a nightmare? It's a nightmare if you're like running a dry cleaner shop. It seems like a much deeper nightmare if you're the police.

• 23:13Speaker 03

  We did this project when I was at the Arnold Foundation where we were trying to change the way sort of lineups are done, you know, photo arrays where a lot of police departments are too small to do sequential photo arrays where you show one picture after the next. And the research is, pretty much falls in favor of that versus looking at six photos or 12 photos at the same time. Because when you look at six photos at the same time, you're sort of thinking like, well, who looks the most like it? Versus you see one picture, you're deciding yes or no. You see one picture, you're deciding yes or no. Most police departments in America are under 10 people. And so we also didn't want the same officer who was making the arrest to be in charge of the photo array. And so to do it, we thought, okay, let's just build an app on an iPad.

• 23:57Speaker 03

  And so we hired a rapid-fire

• 24:01Speaker 03

  Development team in New York, we had a bunch of officers come in who to help us. And literally, the first meeting was the developers were all, as you can imagine, T-shirts, shorts, no shoes. The cops all came in, either fully dressed in uniform or in suits. And this woman working with me was like, why can't they at least wear flip flops of the developers? Like, why?

• 24:24Nilay Patel?

  They actually weren't wearing shoes?

• 24:26Speaker 03

  And it was just, I mean, worlds colliding.

• 24:28Nilay Patel?

  It's amazing.

• 24:29Speaker 03

  Right? It was like, I always keep that in mind. Like, and ultimately, they worked really well together, but it was, it was a long, it was a long process.

• 24:43Speaker 05

  This episode is brought to you by Google Chrome. You think you know a browser, but Gemini and Chrome, that's new. It can help you with practically anything on the web, like restoring a vintage motorcycle from a 50-page restoration block, or finally break down that long article you've had open for weeks. Gemini and Chrome is here for it. Ready to make anything online make sense? There's no place like Chrome. Check responses set up required compatibility and availability varies 18 plus.

• 25:07Speaker 06

  Stonefire Naan is made with real buttermilk and ghee. It's the perfect flatbread for any occasion.

• 25:13Speaker 07

  I put out Stonefire Naan with dips at every dinner party.

• 25:16Speaker 08

  I make naan grilled cheese when my daughter has friends over.

• 25:19Speaker 07

  I use Stonefire Naan to reward the humans who do my bidding. Right. It's perfect for camping trips. Dance for it, feeble mortal. Sorry, is he talking to me or? Dance, ha, ha, ha, ha, ha.

• 25:33Speaker 06

  From our tradition to yours, Stonefire. Together we Naan.

• 25:39Nilay Patel?

  So you brought up CCTV. I want to make sure I call and ask some of these questions because this is very much your zone around data-driven policing. But surveillance is a huge topic in our part of the world. There's both the public surveillance that's going on everywhere. There's data collection. There's like a Verizon NSA building looming in the distance behind us somewhere.

• 25:57Speaker 03

  Watching us.

• 25:58Nilay Patel?

  Yeah, almost certainly watching us at this point in time. And then there's like an explosion in private surveillance, right? I have cameras at my house. There's, I think Motherboard has been an amazing series of articles on ring systems being sold to police departments or police departments selling ring systems. That seems at once great. Like, I bought the cameras for my house. I like having them. They're great.

• 26:21Speaker 03

  Did you buy them for security? Or do you have a baby? Like, we used a camera when our little guy was young.

• 26:26Nilay Patel?

  So we do have a baby. I won't let anything Max does have a user account or a password. So we bought a non Wi-Fi camera for the baby, but for the exterior of our house, we have some cameras. But like, that's a super popular category. Amazon's not selling hundreds of them or thousands of them because people don't like them. People love them. How does that map together? It seems so dangerous to say, "There's a camera everywhere."

• 26:50Speaker 03

  Yeah.

• 26:51Nilay Patel?

  And that it will help the police and make better decisions. We can deploy the police based on what we see on cameras in real time.

• 26:56Speaker 03

  Yeah, I think about this all the time because I think, you know, Colin's asking great questions about these questions of bias in the data. And we all, I think, have the right reaction, which is, we don't want to use data that's biased, or we don't want to have problems. And yet, in our personal lives, we give access to a huge amount of information. And a lot of it is not public. So your cameras, the external area, we could argue is public, but internal is not. But you're still giving access to the companies to have access to your personal data. And it is a little creepy, right? I mean, you know, that's not the technical legal term for it, but there's a way in which, when you think about some of what China's doing with facial recognition and trying to basically, literally, track people in communities,

• 27:42Speaker 03

  you could see where it could end. And one of the things about Ring is that, what has been fascinating to me is that, and some of this is my beef with their marketing, frankly, is like, the marketing around, "Oh, we'll make you safer, and we're going to report everything out to the police departments." It also gives people a sense of being unsafe, in a way that's not true. And so, when you look at the microdata in, for example, New York City, when they do micro polling around, "How do people feel? Do they feel safe in their communities? How do they feel about their police?" It's a fascinating thing, which will not surprise you, which is that people generally say, "Yeah, New York is really safe. My neighborhood is great."

• 28:23Speaker 03

  But the city itself is really dangerous because what leads the evening news, rape, robbery, murder, what do people fear, rape, you know, all of us fear for sort of folks. And I think things like Ring, when they're marketed around crime and reporting back and they're pushing out the crime stats, make, gives sort of people a sense, maybe, that they're not actually as safe as they are, when in fact, you know, I have a beef with the fact that they're four of the most dangerous cities in the world, in America, right? And I have a huge problem with that because I think we actually know how to police cities and we know how to reduce crime significantly, and those are all poor cities who, there's no reason that they're not safe. So I have a beef with that, but overall, we are an extraordinarily safe country.

• 29:07Speaker 03

  And so I think they're sort of marketing off people's fears, and of course, we all fear-

• 29:11Nilay Patel?

  They absolutely are.

• 29:12Speaker 03

  Yeah, and we all fear, right? Nobody wants anything. You have cameras because of safety and security. You want to know who's coming, you know, who's coming near your house, or maybe you want to know who stole your Amazon package, which is apparently a big use right now. It seems to be a very big use of them right now.

• 29:27Nilay Patel?

  No, we have the CEO of Ring has been on the show. It's like Jamie speaks with religious fervor about the mission of his company, which is decreasing crime.

• 29:37Speaker 03

  Yeah.

• 29:38Nilay Patel?

  And he, that's what, at least he's selling that's what he believes. I think it is all but impossible to be that, like aggressive about it unless you actually believe it.

• 29:47Speaker 03

  So that may be his mission, but if we're really honest about it again, I mean, the goal should be to prevent crime. And this is, I would argue, a failing of American policing, is that we've become very reactive, right? Again, it's the 911 call loop, and it's the sort of, somebody reports a crime, you're, even if you capture somebody stealing your Amazon package, that crime has been committed. And so unless your ring is connected directly to the police department, and by the way-

• 30:12Nilay Patel?

  Which seems like the goal.

• 30:13Speaker 03

  Even if they are, I just want you to know this, the officer is not coming to your house. Like, let me be clear. Let's just be totally honest about it, because I'm busy prioritizing where it should be.

• 30:23Nilay Patel?

  It's a really expensive package.

• 30:24Speaker 03

  Okay.

• 30:27Speaker 03

  If you've like, I'm trying to think like, if your cyber truck is stolen from the front of your house, like, okay, but there, the

• 30:35Nilay Patel?

  Wait, you're saying the cops would come and that's it?

• 30:37Speaker 03

  I'm saying they might come.

• 30:38Nilay Patel?

  They'd be like, I want to check out that cyber truck. They need to know where it is.

• 30:41Speaker 03

  So, again, and what the, what the video gives you is the ability to track down the person, in some ways, if you have a known photo for them, right? So, we're already getting into the next layer of how are we using people's photos that, for someone who's already in the criminal justice system. And, you know, I'm, Barry Friedman, my colleague at NYU, writes a lot about facial recognition and the flaws in facial recognition technology. They'll be improved over time, I have no question, but there's still a lot of concerns about using it for law enforcement purposes right now. And so, you know, you sort of end up in the space where I think it makes you feel good, but it's historic, right? There, there'll be a record of somebody having stole your Amazon package, maybe the police department is able to do something, maybe they're not.

• 31:25Speaker 03

  And again, it, it's better than not having any idea of who stole your Amazon package. We've got a lead and we've got something to go on, but the question is, you know, what other data would the company share with the police or with government? And can we limit that? And again, it comes back to this conversation of, we should be having a conversation in every city and nationally about, when we think about government and police departments or government agencies, what data do we use? What data don't we use? What level of transparency is it fair to expect of the police department information they should be putting out, prosecutors' offices, courts? And we should have a conversation now because the private companies, they're not going to wait. I mean, they're, they're doing what they're doing and they're taking your data and, you

• 32:09Speaker 03

  know, I, I don't think they're selling your camera data at this point in time, but

• 32:13Nilay Patel?

  They're certainly sharing it, right? I mean, I, I think that's why that they're so aggressively pursuing the cops.

• 32:18Speaker 03

  Yeah.

• 32:19Nilay Patel?

  And that seems, like, it's happening, the, the way you're talking about how they market it is, we'll put up cameras all your house, someone's stolen your package, the cops will immediately have that video. That's a, they're sharing the data as a selling point.

• 32:31Speaker 03

  Right, yeah.

• 32:32Nilay Patel?

  And that seems extremely disturbing to me.

• 32:34Speaker 03

  Yes.

• 32:35Nilay Patel?

  But at the same time, it's like, well, if someone broke into my house, like, I'd probably share that video.

• 32:39Speaker 03

  Right, and so, look, for, as someone who's run a police department, do I want that data? Oh, yeah. Like, would I like that video? 100%. And so, you know, I would sort of take the view, everything public, everything that's public view, and New York City is a great example. Maybe you could find a couple of blocks that aren't on video camera, but it wouldn't be that easy, right? I mean, between all of the Homeland Security and all the PD stuff, like, it's a pretty wired city. And so, I assume everything I do publicly is public, as should we all. But, there is something a little different about you directly sharing, you know, your sort of, um, let's say you have a fence out in front of your house to your front door, you're sharing that information with the police. Again, as a police officer, I want that information.

• 33:24Speaker 03

  As a citizen, somebody who lives in the community, I think we should have a conversation about what do we share and what do we do with what we share.

• 33:31Speaker 04

  Yeah, I'm really curious, as a police officer, I mean, we hear, I think, a lot from civil libertarians, like you mentioned, about facial recognition, and I'm glad that you brought it up, because, like you mentioned, you know, there's so much data floating around already, we have DMV photos, you know, mug shots, it's very easy for even small departments, soon, it seems like, to just flip the switch and to have facial recognition on. But, there's still a lot of conversation right now about what regulation for that should look like, so I'm just curious, from a law enforcement perspective, what do you think that should look like?

• 34:03Speaker 03

  So, I think you're, I might be wrong on this, I'm not an expert on facial recognition, but I think you might be bullish on a six-person police department having facial recognition capability tomorrow.

• 34:14Speaker 04

  Maybe not like tomorrow, but...

• 34:15Speaker 03

  No, but, at some point, it's possible. And, and if I sit here and I think about it, it will be because a private company sells it to them in a way that is easily, is easy for them to use. And, and then I sort of have this, this point in my mind, which is, who should make that decision? Like, should it be the city of New York? Should it be the state of New York? Should it be the United States government? And that, that is a point that I'm, I'm still thinking about and trying to understand, like, how should we set these boundaries? Should it be local community, where we walk in and say, "Look, we're comfortable, I'm comfortable sharing it with the police department, my Ring video, but I'm not comfortable with it going anywhere other than that, and I'm not comfortable with it being used for anything other than when something has happened that's a negative event, that I want to be able to, like, potential criminal activity report."

• 34:58Speaker 03

  It's sort of like, a direct report. Or, do I, am I comfortable with the police department having full facial recognition? So, I think that's the right question. I also think that, one of the challenges around facial recognition and the amount of attention we spend on it, and I'm not saying we shouldn't be thinking about it from a civil liberties perspective, but I think we spend a lot of time thinking about things when... For my money and my time, I'd rather think about how do we stop people from coming into the system? How do we actually prevent crime? And so facial recognition is not crime preventative at this moment in time. I mean, there are reports that some other countries use it to say, oh, there's a someone suspected of robberies, they're walking in this neighborhood, but we're not gonna, in

• 35:41Speaker 03

  America, we're not going to track people full-time. At least, I hope we never track people full-time. So, again, it's sort of like, how much should our focus and our time be on that? And when police departments want technology, in my experience, they often want it. When you sort of push and say, like, to me, it's never, tech's never the first piece you should start with when you run an operation or you're doing something. The first thing you should ask is, what's the problem you're trying to solve? The second question you should ask is, what's the best way to solve it? And for me, a lot of data and technology can be hugely helpful, especially in the criminal justice space where we don't really understand what's happening and we make a lot of decisions by our gut. But in the facial recognition space, it's really a question for me of, what are we trying to do?

• 36:25Speaker 03

  Are we trying to catch suspects who've, you know, we have a homicide that was committed, we know who, we think we know who we did it, we can't find that person? Is that the purpose? Or is it just to surveil people more generally, which would make us all, I think, uncomfortable? So I think for police departments, they may buy technology than they always buy technology they don't need. But to me, I would, I would really push on this, what's the purpose of it? And does it really actually solve policing problems? And policing problems should be about police officers in communities, understanding what's happening, walking around, you know, interacting with the community, and really understanding what crimes are a problem in the community and thinking about, how do you drive down those crimes?

• 37:05Nilay Patel?

  But let me put you in sort of the opposite role.

• 37:07Speaker 03

  Yeah.

• 37:08Nilay Patel?

  You play the civil libertarian now. And I will try to... Be the over aggressive cop. I'm not saying you're an over aggressive cop.

• 37:14Speaker 03

  I worked in civil rights at DOJ, so I'm good. I can be on either side, right?

• 37:18Nilay Patel?

  It seems like you said, we're not going to track everybody all the time. You said, we're not trying to surveil anybody. It seems like we're just like minutes away from accidentally combining four or five public and private data sets, and then just surveilling everybody all the time anyway.

• 37:32Speaker 03

  Right.

• 37:32Nilay Patel?

  Right? Like, I can't make transactions in New York without generating a record that goes to Visa or MasterCard. I can't walk around New York without being on a bunch of cameras, public or private. Maybe facial recognition is not easy except now. It's pretty easy. It's going to get a lot easier.

• 37:49Speaker 03

  I agree.

• 37:49Nilay Patel?

  Right? Like, Google Photos will just like happily recognize faces in a data set today. Facebook will do it. If they're deploying it for consumer ad-based services.

• 37:56Speaker 03

  Yeah.

• 37:57Nilay Patel?

  Someone is going to seize that opportunity.

• 37:58Speaker 03

  Yeah, I agree with that.

• 37:59Nilay Patel?

  Amazon is happily thinking about doing stuff for, well, they're suing the government because they didn't get the contract they wanted, but they're happily thinking about working with the military and police. So, the capabilities are there, and I agree with you. You don't think with the tools first. But it seems almost like unless there's proactive regulation, and this is what I'm asking you to be in the other seat, unless there's proactive regulation, the cops are just going to do it by accident. It's going to get there way faster than they anticipated.

• 38:23Speaker 03

  So, some stuff may already be happening by accident already. Again, I'm not attributing ill motives. I'm just saying, you and I and all three of us probably have seen when you have vast data sets and a lot of information, who gets access and how people access it, and sometimes they do access things that they don't mean to access. I've had people send me data sets with social security numbers and identifying information and, you know, personal identification information, and I, my head explodes, right? And I, take that, take that back.

• 38:50Nilay Patel?

  My favorite example is just when people are tweeting pictures of their boarding passes for airlines, and it's like, stop!

• 38:55Speaker 03

  I know.

• 38:57Nilay Patel?

  What are you doing?

• 38:57Speaker 03

  Don't do it. I know, right? But again, people don't think, right? And so, I think that you're right. There's some accidentally happening, and there could be some intentionally done. So, I'm not in any way saying that, when I'm saying, we don't want to do it, I'm saying, we should make sure it doesn't happen, and we should make sure that there are, here's how I would frame it. There are legitimate law enforcement reasons why I want access to the cameras on Broad Street. And I just want to know if a crime's been committed, and again, a lot of those CCTV cameras can be moved, their ShotSpotter technology, a gunshot is fired, it tries to sort of video that incident if somebody's running away from the scene and try to capture information.

• 39:40Speaker 03

  And again, those are just leads that help you solve crimes. So there's legitimate law enforcement purposes that should, we should require that they be articulated, and we should require, again, what are the purposes of using, like, I could articulate why you need a camera on Broad Street, it's also publicly available space, but we should be having this conversation about when it's okay to use it, when it's not okay to use it. And we also have to be careful, because I think government, and I've worked in government for a long time, and, you know, here's the setup for "I love government, but..." But it's not, it's not iterative. And so, one of the, one of the things I see with technology is that the tech is changing so rapidly.

• 40:22Speaker 03

  So you're right, it could be a year, we could be a year away from valid facial recognition technology, and if we make rules today, will that be applicable with the tech of a year from now, will it be applicable in five years? And so, we have to figure out, and do we create a separate government agency, is there like, you know, the Federal Election Commission, is there a group that we put together that's responsive on a regular basis to these questions? But yes, facial recognition is a great example of a space where unwittingly, we could all of a sudden be in a space where the government is literally tracking every move. And, you know, again, it's one thing for the police to be able to look at a camera and evaluate footage to see, okay, a crime was just committed at that location, there's

• 41:05Speaker 03

  a reason I'm watching that camera, or a crime was committed yesterday, let me see, you know, is there a lot of foot traffic, what's happening? That's a legitimate purpose of looking at it, versus, I'm going to knit together your ring with the CCTV on the street, with your Google transactions, with your Apple Pay, and I'm going to literally map out your entire life. And, and it's not just a question of government doing it, it's also a question of people who commit crimes doing it. Right? So we should be wary of the technology being used for, not just by government for surveillance, but also by people being able to hack into, if those systems get integrated, essentially hack into your entire life.

• 41:40Nilay Patel?

  So Colin, you've covered the tech industry sort of asking for this regulation. Does that map to what Anne's saying over here?

• 41:47Speaker 04

  I think so. I mean, you know, it's interesting. I think you saw Microsoft come out and say, "Give us a law. We want something." But it does seem like companies are narrowing exactly what they want. You know, they're not saying they clearly don't want just a blanket ban on facial recognition, which it seems like some places want. But at the same time, I think we're seeing police officers coming out and saying, "No, like, we don't essentially want any ban on this at all." And so it's sort of like, I don't know, the back and forth has been really, it seems like we're looking at two extreme ends of it right now.

• 42:22Speaker 03

  Yeah, and probably neither extreme is right, in my view. That I think there could be legitimate law enforcement purposes, but again, I think they have to be articulated. And there's value in having a community conversation about what are we comfortable with law enforcement having access to and when. I also, not to be cynical for a moment, but let me bring a little, let me bring a little cynicism into the room. The tech companies are asking for federal government regulation because what they don't want are AGs, like I used to be, saying, "Hey, we need a state law." And then there are 50 state laws, and it's a mess of a regulatory scheme, which, to be fair to them, would be complicated, especially for startups, if you're trying to run a company and you're, you know, you don't want

• 43:05Speaker 03

  to have 50 different rules of when things could be used and when they can't be. So, but there is, there is like, there's an incentive for them to basically say, "Hey, give us a very minimal federal law that doesn't really stop us from doing anything, but puts up a little bit of a guardrail, and then that lets us operate in all 50 states." And so I think that's a little bit of what we're seeing. You know, on the tech regulation question, I think it's a tough question, and facial recognition is just one example of it, where... To me, I mean, the amazing thing about the tech industry is the creativity and innovation, which we do not want to stop. I think we've also seen that there are places where we need guardrails for American society. Otherwise, we won't, you know, have confidence that our best interests are at heart and that, you know, there's abuses that can happen.

• 43:50Speaker 03

  And so it's going to be really complicated to figure out, like, what are those guardrails that allow companies to retain sort of innovation and the ability to do things that may be transformative without, you know, really hurting Americans, whether it's privacy or, you know, we've seen it with political ads. I mean, there are so many examples right now, I think, of spaces where we have to be worried.

• 44:13Speaker 04

  One reason I think that the private sector has, you know, so eager to sell these sorts of tools to cops is because government, the aisle of government, but they're not always so great at accurately collecting that data. So I'm thinking of, like, Chicago Police has this sort of infamous gang database that a lot of people say has, is full of errors. I mean, how do you make sure that when government is collecting this data, that it's auditable, that it's accurate, that what you're collecting is true? Because the risks are so high if it isn't.

• 44:43Speaker 03

  Yeah, again, I mean, sometimes I think, this is true generally in criminal justice, is just sort of asking questions about the process that we use to do things. And we look a lot at the outcomes, I think, in criminal justice, but I don't think we spend enough time looking at, like, what is the process by which the police department in Chicago classifies somebody as being a gang member? Wearing a red shirt, not enough, in my view, right? And so, you know, there are departments that basically say, you're wearing a red shirt, you belong to one gang, you belong to a gang. That's such a low standard. And to me, if you're going to charge someone with a crime, you know, we charge people with crimes based on gang affiliation. Sometimes there's a gang enhancement. If you're going to segregate somebody in detention because of a gang, like, you're making really significant decisions.

• 45:27Speaker 03

  And the question is, what level of proof do we need to have? And again, some of this stuff just happens organically in police departments who say, oh, we should make a list of the gang members. And they're doing it through anecdote, or they're doing it through... You know, this guy had a tattoo on his forearm, but the tattoo doesn't even get photographed. So there's no real record. And that makes me uncomfortable without really being really rigorous in what's the process that we use. Now, again, understanding what gang someone's in can be incredibly important when you're trying to prevent crime. An example I'll give you is one of the things that hospitals frankly have often been better at than police departments is someone is shot in a community, the most likely person

• 46:09Speaker 03

  to fire a gun is that person, oftentimes because it could be retaliation or somebody close to that person who's going to retaliate if it's a gang war or if it's a turf battle. And so hospitals have long gone to every gunshot victim, not every hospital, but a lot of hospitals go and say, hey, you know, what happened? Who shot you? What's going on? And they're sort of, they're seeing that the likelihood is that they're going to get another gunshot victim or a homicide coming. And police departments haven't historically done that, right? We haven't sort of understood that there's real information. And so if I don't know that you're in a gang, I don't know who your rivals are, I don't know like what I should be watching for. And when you think about prevention, and this is all really sensitive data and really sensitive ways to think about it, there is a certain amount of information that becomes

• 46:53Speaker 03

  critical for law enforcement to have and use. But the process by which you classify people or understand it has to be a pretty high threshold. And so, you know, I think the other piece, just going back to the tech piece for a second, is that we have not funded or valued data in police departments. And we should be really honest about the fact that we have not, like when I wanted to hire data analysts in police departments, the unions were not fans of mine, right? Because they basically, what did they want? They want a uniform. They want somebody else who's going to be a member of their union, and the officers always want more officers. But that model to me is really arcane, and it's really old school to just think that the way we police is by just having people either on the street or answering 911 calls

• 47:36Speaker 03

  versus somebody actually saying, well, here's the robbery pattern that's happening, or here's the two gangs that are at war. This is why so many people are being killed. And so. We have to, sort of, in some ways, if we train and regulate and enable police departments to do their own data work, I think we may end up having better consumers and then they'll buy better technology instead of looking at the technology, like, being in a space where they are wearing their uniforms and the tech people are wearing no shoes and then, like, the tech people are just coming in and building whatever the tech people think they want.

• 48:07Nilay Patel?

  The union's like, "You got to wear shoes, man."

• 48:08Speaker 03

  Yeah.

• 48:08Nilay Patel?

  "I can't wait to start Police Tech Review." That's gonna be the next thing that we do on the side. Kick off our shoes and... Uh, yeah. We only have a few minutes left with you, and so I just want to stay on, sort of, the cynicism beat for a second. It is impeach...

• 48:21Speaker 03

  Because it's holiday time.

• 48:22Nilay Patel?

  It's holiday time. Today's the day that the House will almost certainly vote to impeach Donald Trump. It's gonna happen, we think.

• 48:28Speaker 03

  Yes, it's gonna happen.

• 48:29Nilay Patel?

  It's definitely gonna happen. Hours from now, when we're recording. This episode will come out later. You, the listener, will know whether Donald Trump has been impeached or not. We do not. We've talked a lot about how we need to rethink regulatory approaches. We've talked a lot about tech companies wanting federal laws versus state laws. We've talked a lot about how we might think about data collection broadly. We've talked about the role of police in our society. We have a totally dysfunctional federal government. How on earth do you square those things?

• 48:55Speaker 03

  That's a big question for our last two minutes.

• 48:58Nilay Patel?

  You've got three minutes, but, you know, take a stab at it.

• 49:01Speaker 03

  Yeah.

• 49:02Nilay Patel?

  Also, you have a whole podcast that's, like, fundamentally about this.

• 49:04Speaker 03

  Yes, and we, Preet and I, talk a lot about impeachment and a lot about the rule of law. And the thing that I think, I would say, troubles me the most, or the thing that I worry about the most, is that the institutions and the rule of law matter to me, and the Department of Justice is a great example of men and women who, in my experience, I was at DOJ, I've worked closely with the FBI and the other DOJ entities, just being nonpartisan and being, sort of, not an arm of the president matters a lot. And I don't think we're seeing that right now. And so, we saw this letter from the president yesterday, which is... I mean, you know, I've read it and then I read it again and I thought, first of all, how could he have written it? And second of all, how could anyone have let him send

• 49:44Speaker 03

  it? You know, we're, we're in this territory that's sort of uncharted. And if I put it really bluntly, the, the thing that I think we're at risk of is losing our, our sort of institutions and our constitution. And here's why. The president makes a call, in my view, it is based on the evidence we're seeing, it is worthy of impeachment. And so people can differ about this, but there's, in my view, a considerable amount of evidence that the president did abuse his power. And the second part is almost the more important part, which is that he categorically denied access to evidence, to witnesses, with this argument that it wasn't legitimate. And the problem in, for all of us, and we should all be, I think, deeply concerned about this, is that the constitution holds the president

• 50:28Speaker 03

  accountable to Congress and to the courts. There are three co-equal branches. And the president doesn't get to decide whether or not a congressional investigation is legitimate or not. And what the president has done here is said, no, I'm not answering you. Right? I mean, imagine if you say to your child, like, you know, you have to stop throwing your food. And your kid's like, no, I don't answer to you. Right? Like, you're not the boss of me. Right? It's possible. But, you know, it's like the president being like, I, I'm not going to hear you. And then the Senate not requiring evidence and testimony is a rubber stamp of the president's sort of categorical denial of the authority of Congress. It really does mean that the president, at the end of the day, can kind of do anything. If he finds that a congressional inquiry is illegitimate, it gives him enormous power

• 51:13Speaker 03

  to then basically just say, I'm not going to answer you, and I'm not going to present witnesses or documents. And that institutionally worries me enormously. The importance of the police, the importance of the Department of Justice, the importance of the FBI, they're all flawed. I mean, you know, nobody thinks that these places are perfect, or all the decisions made are right. But there's a certain basic sense, I think, we have of them being, people make the best decisions they can for the right reasons. And that feels to me very much at jeopardy. So I think I'm, I'm ending on cynicism, which seems bad.

• 51:44Nilay Patel?

  Happy holidays.

• 51:46Speaker 03

  Happy New Year.

• 51:47Nilay Patel?

  We'll try again next year. Everybody take a couple of weeks off. How do you think that's going to impact this broader conversation we've been having? Because I think just in the last few weeks, we've seen California has passed a privacy law, AB5 has gone in to sort of regulate the gig economy. You are seeing this sort of state by state. There's a net neutrality debate that we cover all the time that is happening now at the state level versus the federal level. Do you think it's going to actually shift in that direction?

• 52:11Speaker 03

  So I think states are really important and I think, yeah. And I think people under, I know, that could be the quote of the day. States are really important. But I think people undervalue just how much the states can and will do. And when you have a president like Donald Trump, who, and by the way, it was true of Obama too, Greg Abbott was the Texas AG at the time, and he was a Republican, and he was quoted once as saying, "I get up, I go to work, I see the president, I go home. The next day, I get up, I go to work." And we're seeing that a lot with the state, the Democratic state AGs now. But we're seeing this in environmental, we're seeing it in privacy, we're seeing it in a host of areas where people don't think the federal government is acting in sufficient ways.

• 52:52Speaker 03

  And it's almost like, are we going to get to a point where, you know, you could move states, but if there's not federal action on the environment, it's not going to, right, it's not going to help us that much. So I think, I think state action is really, it's under, it's underappreciated how important it is. But it's also because we're 50 states and because so many things cross state lines, it's really, really critical for the federal government to be engaged on this stuff. Just to talk about criminal justice reform for a minute, I think what concerns me about it is that there's two things that concern me about it, and the sort of movement in general. And again, I believe there should be less crime and less incarceration.

• 53:32Speaker 03

  I think there's a really critical public safety aspect to not losing as we, we should dramatically reduce incarceration. And again, I think in my lifetime that will happen. But we have to also understand that there are a small but critical group of people who do really pose risks to public safety. And I think being just really honest about that, and my view is working as hard as we can to prevent people from getting to that point and getting to a position where somebody is actually a risk to public safety, is very important. We've seen a couple of high-profile crimes recently, most recently the murder of the Barnard student. I worry a lot about the kickback, like the kickback to reform movements where there's an argument that we have to be more aggressive.

• 54:13Speaker 03

  You saw the Sargent Union basically coming out and criticizing the young woman and criticizing the mayor. And look, I'm a huge critic of the mayor. I think he's not my choice, right, for mayor, for a variety of reasons. But I also think that when you think about crime, horrific crimes like that cause people to react with fear and oftentimes to basically say we need a stronger law enforcement response, which is just the law and order response. We arrest everyone and we incarcerate everyone. And so what I worry a lot about in this world is that we be really smart about how we think about reforming institutions and we don't forget that, to me, the goal is always public safety. We can be wildly better at how we do that, but we have to really be thoughtful about it.

• 54:56Speaker 03

  And we have to really sort of figure out how do we neutralize the fear that people in communities legitimately experience and acknowledge it and recognize it without saying that that means we need more incarceration. And so, and by the way, I'm not saying I don't want anyone to think that I don't believe that the young men who committed those crimes deserve to be incarcerated. That's, you know, I do, right? What I'm more saying is that the response overall, it can really be blunt force. And we've seen again and again in criminal justice where we just increase sentences because we've seen a horrific crime, where we create a new crime. And so I think we have to figure out how, as we move into this point in time, like how are we really smart about policing America in a way that makes people safer, prevents

• 55:39Speaker 03

  crime as much as possible, and really like honors the sense of justice and fairness in a community.

• 55:44Nilay Patel?

  All right. Well, we've taken up so much of your time. Thank you for joining us. Where can people find you?

• 55:48Speaker 03

  At NYU, mostly. You want my Twitter handle?

• 55:51Nilay Patel?

  Yeah, Twitter and podcast. I was going for that.

• 55:53Speaker 03

  Okay, yeah, sorry.

• 55:54Nilay Patel?

  Or just walk the campus.

• 55:55Speaker 03

  [laughter]

• 55:56Nilay Patel?

  You show up and ring cameras around campus.

• 55:58Speaker 03

  Cafe Insider, Preet and I do a weekly podcast. And then on Twitter, I'm just at Anne Milgram.

• 56:03Nilay Patel?

  Great.

• 56:04Speaker 03

  I think I'm on Instagram too, but I have no idea what that is.

• 56:06Nilay Patel?

  That's good. That's for the best for everybody. It's for your own mental health for everybody. All right. Thank you so much, Anne Milgram.

• 56:11Speaker 03

  Thank you. Thanks, guys.

• 56:13Nilay Patel?

  All right. Thanks to Anne Milgram for joining us. You can find her on her show, Stay Tuned with Preet. It's a great show. It's out in the world. Go listen to it. We'll be back later this week with the chat show. Back in with the interview show on Tuesday. On and on we go. Love hearing from you. You can tweet at me. I'm at Reckless. I'll talk to you soon.

• 56:33Speaker 09

  Everyone's got a hard question about AI. Your job, your kids, where it's going. Anthropic was built to surface those questions and share what it finds along the way. Because there's hope in hard questions. Ask yours at claud.ai/spotify and keep thinking. That's claud.ai/spotify.

• 56:56Speaker 06

  Stonefire naan is made with real buttermilk and ghee. It's the perfect flatbread for any occasion.

• 57:02Speaker 10

  I put out Stonefire naan with dips at every dinner party.

• 57:05Speaker 08

  I make naan grilled cheese when my daughter has friends over.

• 57:08Speaker 11

  I use Stonefire naan to reward the humans who do my bidding.

• 57:11Speaker 07

  Right. It's perfect for camping trips. Dance for it, feeble mortal. Sorry, is he talking to me or? Dance.

• 57:22Speaker 06

  From our tradition to yours, Stonefire.

• 57:24Speaker 07

  Together, we're naan.