But what happens when a police force or a single police officer can fire up an army of a thousand AI agents to do this kind of long form, what would normally take a month of research and work, what can go wrong with that?

Sure.

I mean, I think the promise of AI agents is to create the answer machine for police investigators, right? So if you think about how AI works and the way you use it almost every day, it used to be when you, like, Googled something, you were searching for something.

You got a whole bunch of links that gave you a way of finding the information.

And now when you Google something, you get the answer, right? So the same thing happens in policing, right? So if the police have all those databases we were talking about, ALPR datas, video databases, uh, people's addresses, nine one one calls, gang members, all the social networking connections that Garance was talking about, it used to be that an officer would have to track down each one of those clues and sort of connect the dots.

In a world of agentic policing, you ask for the answer.

So take a concrete example, like you have a carjacking, you have a gray car, and you have a partial license plate, right? So you gotta figure that out.

So Andrew, I wanted to start with you.

Um, does the good outweigh the bad? Are there solutions for the problems we're facing right now? Andrew Ferguson

FergusonThat's actually why I wrote this book.

It, it's about how we have, uh, developed systems of surveillance, whether we're talking about sort of democratically mediated systems of surveillance like, uh, automated license plate readers like Flock or real-time crime centers with thousands of video cameras recording the streets and being able to track people, uh, where they go.

Uh, we have developed a world where we are watching ourselves in a new way, where we've kind of flipped the default.

Right now, everyone is under surveillance all the time.

It used to be the case that you were only under surveillance if you kinda did something to draw the attention of the authorities.

Uh, and then we had systems like warrants and, and, and individualized suspicion that would cause people.

Yeah.

Um, does the good outweigh the bad? Are there solutions for the problems we're facing right now? Andrew Ferguson.

That's actually why I wrote this book.

It, it's about how we have, uh, developed systems of surveillance, whether we're talking about sort of democratically mediated systems of surveillance like, uh, automated license plate readers like Flock or real-time crime centers with thousands of video cameras recording the streets and being able to track people, uh, where they go.

Uh, we have developed a world where we are watching ourselves in a new way, where we've kind of flipped the default.

Right now, everyone is under surveillance all the time.

It used to be the case that you were only under surveillance if you kinda did something to draw the attention of the authorities, uh, and then we had systems like warrants and, and, and individualized suspicion that would cause people.

And we've kinda reversed that, that default, where we make it, uh, uh, all too, uh, easy.

Yeah

What are some of the other questions that this ruling leaves open for the future when it comes to protecting digital privacy?

So we know now that we, citizens and people in the United States, have a reasonable expectation of privacy in our location data.

And that is clear in our cell phones because that was the facts of the Chachere case.

It's probably pretty clear in our apps and the digital apps we have.

Chhatra clarified that our cell signals, like our cell site location information, like how we connect to cell towers, is pretty much protected.

And so things like tower dumps from cell phones, where basically police are collecting all the phones from a particular cell tower, are now protected.

The open questions, I think, are going to be about systems of surveillance.

And how does AI sort of fit into this picture?

What are some of the other questions that this ruling leaves open for the future when it comes to protecting digital privacy?

So we know now that we citizens and people in the United States have a reasonable expectation of privacy in our location data.

And that is clear in our cell phones because that was the facts of the Chachere case.

It's probably pretty clear in our apps and the digital apps we have.

Chhatra clarified that our cell signals, like our cell site location information, like how we connect to cell towers, is pretty much protected.

And so things like tower dumps from cell phones, where basically police are collecting all the phones from a particular cell tower, are now protected.

The open questions, I think, are going to be about systems of surveillance.

And how does AI sort of fit into this picture?

With my grandparents, how would I know?

Or how many of you, like, literally, you know, Googled and wrote down the directions? We used to have these things where you would have to, like, go and you'd print out the directions and write down MapQuest, right? Or even the Triptix, if you're really old, you know, the AAA, and they'd flip things so you could go from place to place.

And, of course, no student raised their hand when I asked them questions.

And I'd say, well, how many of you use your phone or the GPS in your car? And, of course, everyone said, yes, we do.

And then the follow-up is, how many of you know that that data is available for law enforcement if they want to get access? And everyone's like...

Sort of, yeah, I don't really know.

And I said, well, how many of you will change your behavior? And, of course, no one will.

What does that mass surveillance mean? Who or what is watching?

Why, why is it such an, an easy ask?

So the first thing is that the standard is probable cause.

Now, probable cause is in the Constitution, so that's the standard, but it's pretty low.

So 51%, uh, is a preponderance of evidence, and probable cause is decidedly lower than that.

Uh, that means that police can go there, uh, with, uh, information that could be wrong.

It just could not be that persuasive.

Uh, and that opens the door to all of your digital information if they have probable cause of a crime.

[laughs]

What are you seeing right now?

One of the real concerns with AI police reports is it is the document that police have to rely on, obviously, prosecutors have to rely on, judges have to rely on in a preventive detention system where they're given the affidavit.

And in a case, uh, of our current criminal justice system where 95% of cases result in plea bargains, the police report almost never gets interrogated.

So here you have this, like, document that we all rely on.

I was a public defender before teaching.

Everyone relied on the police depart- police report.

But now if that has been AI-generated or partially AI-generated, it means the entire system is sort of, like, infected with this one document that's controlling the whole pathway of a criminal case in ways we haven't fully comprehended or, or thought about, well, what does it mean? And that might not be the fault of the technology.

But from your perspective, how much should efficiency drive the design of the criminal justice system?

For example, yesterday, I d- maybe because I, I had this conversation with you on the brain in anticipation, I took note of something that perhaps I would've otherwise ignored, and that was I was driving in my community, and there was a median strip, and a police car was parked right in the center of it, kind of an odd positioning, and as I drove by, I said, "Probably using license plate readers."

Unquestionably.

Almost all, uh, police cars now have some form of license plate reader.

All the meter people who go by and give tickets also have it.

It's on street corners.

Uh, we also have video cameras, right? So we're talking about one of the tools in the police, uh, arsenal toolbox.

We also have, uh, video systems that are connecting, uh, to real-time crime centers.

Your thoughts?

Yeah.

Uh, but it basically says whether or not something is a, quote-unquote, "search" for Fourth Amendment purposes, and again, we're talking about unreasonable searches, is whether or not it violates an expectation of privacy.

So easy example, your home.

Your home is a place of, of privacy.

You live with your family, you live by yourself, and things that you do in your home generally cannot be seen by law enforcement.

And it is pretty much accepted that in order for police to go into your home, they need to have a judicial warrant, uh, because you have a reasonable expectation of privacy in your home.

The hard question is, what if you have cameras in your home that don't just stay in your home, but are also going to a third-party cloud provider like Google or Amazon? What happens if you have microphones in your home called Echo devices that are going to Amazon? What if the police wanna get access from things in your home, but they're not going to your home, they're just going to the third party, the private company you asked to, uh, hold this data for you? Do they need a warrant? It's an open question.

Yeah.