
The Golden State's AI gamble.
Eoin Hinchy: It sets out to categorize different types of risk and systems of risk where AI will be used, everything from unacceptable risk right through to kind of like low risk, and the various guardrails that should be applied across those -- across those different categories.
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 co-host Ben Yelin from the University of Maryland Center for Health and Homeland Security. Hey there, Ben.
Ben Yelin: Hello, Dave.
Dave Bittner: On today's show, Ben looks into the doctrine of preemption and how it might affect efforts to regulate data privacy. I look at a major grocery chain getting pushback on its facial recognition program. And later in the show, Ben's conversation with Eoin Hinchy, Tines' co-founder and CEO, sharing his perspective about straddling EU and U.S. regulations. 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. [ Music ] All right, Ben, we've got some good stories to share this week. You want to start things off for us here?
Ben Yelin: So who's ready to talk about the doctrine of preemption? Can seem --
Dave Bittner: Oh, I'm sitting on the edge of my seat.
Ben Yelin: I can see everybody getting super excited. Even when I mention this to law students, I think some eyes start to roll.
Dave Bittner: Uh-huh.
Ben Yelin: But I noticed something really interesting. I first came across it on the Electronic Frontier Foundation website, and that linked to an op-ed by a guy by the name of Jake Snow on Tech Policy Press. Jake Snow is a senior staff attorney at the ACLU of Northern California -- go NorCal. And he wrote an article about what he sees as big tech's effort, similar to the effort of tobacco companies in previous generations to use the doctrine of preemption against consumers, and I'll explain how he thinks that's going to happen. The ACLU and a bunch of other groups also wrote a letter to Congress about this issue. The way preemption works is Congress can write into a piece of legislation that a bill explicitly preempts state law on a matter. So Congress could write this bill explicitly preempts all state laws relating to, for example, data privacy legislation or certain aspects of data privacy legislation. There can also be implied preemption if Congress has indicated the desire to occupy the field. So even if they don't explicitly preempt state legislation, if there's something inherent in the statute that makes it clear that Congress wants to have full dominion over this topic, then courts could recognize implied preemption, and the federal law would still preempt state laws.
Dave Bittner: Can you give us some examples, or are there well-known examples of this?
Ben Yelin: Yeah, I mean, we see things -- a lot of things related to like federal highway administration superseding state law -- anything that has to do with interstate commerce, those are areas that are federal domains, and they're federal domains because those are explicit federal powers listed in Article I, Section 8 of the Constitution. So there are lots of items. Basically, anything that involves regulation of interstate commerce could fall under preemption. People have studied it a lot on the question of marijuana legalization, an area of interest, and also on abortion, now that we have the Dobbs decision -- and that I will be discussing a little bit later, as I think it's an interesting element to all of this.
Dave Bittner: Okay.
Ben Yelin: So the reason this is relevant to data privacy is states like California have passed really strong legislation providing for robust protections for consumers as it relates to their personal data. Obviously, California is at the forefront of this. They're not the only state that's done so. Several other states, including now Maryland, our home state; Virginia; and others, have passed data privacy legislation. There has been an effort over the past two sessions of Congress to pass a federal data privacy law, and one of the provisions in that proposed law would preempt state law, and therefore all of the stronger California provisions would be superseded by weaker federal provisions. And the allegation from the ACLU and others is this is part of a coordinated strategy to weaken data privacy protections across the country. And in this op-ed that Mr. Jake Snow writes, he goes through kind of the three-step process of how big tech is mirroring tobacco companies in trying to weave this preemption thread.
Dave Bittner: Yeah.
Ben Yelin: So step one is --
Dave Bittner: Well, wait, wait, wait --
Ben Yelin: Yeah?
Dave Bittner: -- before we get there --
Ben Yelin: Yep.
Dave Bittner: -- let me -- let me -- let me interrupt.
Ben Yelin: I will slow down.
Dave Bittner: No, no. Well, for those of -- for those in the audience who may be too young to have -- to know the history of big tobacco, and I'll just, as a quick aside here, I'm reminded of a conversation I had with my two kids only a couple years ago when they were both teenagers, and they said, "So you were allowed to smoke like anywhere?"
Ben Yelin: Oh yeah, I think people don't realize --
Dave Bittner: In the mall, at the grocery store, in a restaurant, in an airplane?
Ben Yelin: In --
Dave Bittner: Right?
Ben Yelin: -- the airplane is the craziest one to me --
Dave Bittner: Right?
Ben Yelin: -- because it's like a not -- a, yeah, I mean, I think we've made this metaphor --
Dave Bittner: Right.
Ben Yelin: -- the non-smoking section of the airplane is like the non-peeing section in the pool.
Dave Bittner: Exactly.
Ben Yelin: But yeah, I mean, people don't realize how ubiquitous cigarette smoking was --
Dave Bittner: Right.
Ben Yelin: -- even 30, 40 years ago. I'm not that old. I'm sorry to --
Dave Bittner: Well, I am.
Ben Yelin: -- yeah, compare myself to you, Dave, on the -- on this issue --
Dave Bittner: Right.
Ben Yelin: -- but I remember cigarettes everywhere, smoking, non-smoking sections in restaurants.
Dave Bittner: Right.
Ben Yelin: So it's just we have gone through a sea change. I don't think people -- our Gen Z listeners really understand how cigarette smoke just used to be everywhere.
Dave Bittner: Right, and so the tobacco company was very successful in their campaigns trying to, I guess, inject doubt into the notion that smoking was addictive and bad for you.
Ben Yelin: Yeah, and it worked for a while.
Dave Bittner: Right.
Ben Yelin: But then it stopped working, and that's kind of what I think we need to look out for here as it comes to data -- as it relates to data privacy.
Dave Bittner: Okay.
Ben Yelin: So the way this sort of deception works, according to this op-ed, is step one, you have a PR crisis. So for the tobacco industry, it was all of these scientific studies showing that tobacco was very dangerous, that smoking gives you lung cancer, and that puts companies on the defensive. So for big tech companies, it's all of the terrible things that we cover on this podcast as companies aren't able to protect people's data, so companies preying on senior citizens, financially vulnerable. The fact that the lack of data privacy legislation supercharges police surveillance with -- through things like data brokers, how it helps empower the federal deportation system. So this is kind of a similar PR for big tech companies. That's shown up in polling data. Confidence in our big tech oligarchs has decreased significantly over the past several years, I think largely in response to these perceived privacy violations.
Dave Bittner: Okay.
Ben Yelin: So what big tech has done, and what the tobacco companies used to do, is pass kind of targeted piecemeal bills. So what this op-ed writer would call ineffective alternatives only pretended to address the issue. So when we're talking about historically with smoking, it would be non-smoking sections in restaurants, which of course continued to allow smoke and filled the air with secondhand smoke. This happened in 17 states. By 1995, as he writes, Virginia, for example, which is a big tobacco-producing state, passed the deceptively named Indoor Clean Air Act, which was intended to extinguish the growing grassroots energy around anti-smoking bans. So it was these sort of half measures that weren't actually going to achieve the ultimate goal, which was to protect our health and safety. So the same thing is kind of happening now. Big tech has escalated its lobbying, especially after California passed the CCPA. They spent $70 million in 2021. There has been a coordinated nationwide campaign by big tech to mold the rules to its will, and they referenced numerous studies that have shown that states have increasingly passed industry-favorable privacy laws. So that's step one is you manage the PR crisis by acting like you're being helpful and proposing these kind of piecemeal measures that's not going to destroy the industry, but hopefully it'll make consumers happy. Step two is to point to the patchwork.
Dave Bittner: Can I -- can I just --
Ben Yelin: Yeah, yeah, yeah.
Dave Bittner: Let me dig in here for a second because -- and I -- maybe -- I don't mean to put you on the spot here, but I know as part of your work at the University of Maryland Center for Health and Homeland Security, you work with some of our state legislators on these sorts of policy issues, right?
Ben Yelin: Right.
Dave Bittner: Like you share your expertise with them. What is your experience here in the great state of Maryland about the influence of tech companies in trying to have their hand in what we do?
Ben Yelin: I mean, I think it's ever present. I think members of the Maryland General Assembly try to be as aggressive as possible in regulating big tech companies' data privacy. But I've been at these hearings. They always come. They're always very well-dressed. They always have very talented lobbyists, and they make very compelling arguments, frankly, that this is going to stifle innovation. This is going to add to a confusing patchwork of laws. We're going to have to pull out of -- pull our services out of Maryland if you are going to pass such an onerous regulation on us, so it's an ever-present concern. And like I said, even somebody who is not predisposed to agree with big tech companies, they can be very persuasive. So I, you know, I don't think it's one of those situations where lawmakers are just rolling their eyes and understand that this is all about the money. I think they actually make good arguments.
Dave Bittner: Okay, all right. So step two.
Ben Yelin: Yeah, step two in this process is to point to the patchwork. Basically, the way this works is big tech companies now and tobacco companies in the past say, Look at all these disparate state laws. It's so hard to comply with all of them. Some of them are contradictory. We don't want to have 50 state standards. It's just too difficult. And we've seen that, and again, that is a perfectly reasonable argument. It is difficult to comply with 50 separate regulatory regimes. So then they go the preemption route and say, Why don't we have a uniform national standard? And once we establish that uniform national standard, we can use this doctrine of preemption to get rid of state laws that we don't like. And what advocacy groups have been concerned about is that the National Data Privacy Bill would set a floor below the level of protection that exists in a state like California and would prevent states like California from instituting stronger provisions like those that are in the CCPA. I think that's actually what killed the federal data privacy legislation at the end of the last Congress because the Speaker of the House at the time, Nancy Pelosi, was concerned that the bill, as proposed, was going to preempt California state law. And I think that's one of the reasons that data privacy legislation has not advanced this session. It's this kind of backhanded strategy to first encourage a piecemeal approach to trying to -- try to address the PR crisis, work with state legislatures to enact kind of the minimal policy possible to address consumers' concerns, then point to that patchwork and say we need a national standard, then come up with a national standard that preempts state law. This is maybe a little more conspiratorial than I would be in analyzing this problem.
Dave Bittner: Well, yeah, but because I'm -- the question I ask is What's the alternative process?
Ben Yelin: So I think the alternative process, in their view, is having a federal floor in these data privacy legislation, in these data privacy bills, and then allowing states to exceed that floor should they choose. So there would no -- there would be no express or implied preemption. Big tech companies would be against that because they might not agree to federal legislation if they also risk having states go above and beyond whatever the federal government comes up with. But I think, in the eyes of advocates, that would be the most fair solution. As you said, a federal data privacy standard -- that standard is the floor and not the ceiling. States are free to regulate beyond that floor. I said I was going to mention abortion again because this is strikingly similar to a piece of legislation that's been circulating in Congress in reaction to the Dobbs decision proposed by Senator Lindsey Graham, which would set a national standard for abortion, making it criminal at 15 weeks of gestation, but states would be allowed to pass regulations that go beyond that standard. So they could pass six-week bans or zero-week bans, where abortion is banned entirely.
Dave Bittner: But they couldn't pass 20-week bans?
Ben Yelin: Exactly, because you'd have that federal floor, but not a federal ceiling. And I think that's kind -- I realize the irony of this, but I think that's kind of what the ACLU is advocating for here is you set that national floor, and then you try and get the states to work to build more robust protections over and above that floor. It's a really complicated issue because I don't think big tech will be happy with still having a patchwork of laws to comply with, and if they withhold their support from a federal data privacy bill because there's no preemption, that does not bode well for a federal data privacy bill. So there has to be some way to work through this. I don't know what the ultimate solution is, but I really understand the concern that the ACLU is expressing in this letter about the perils of preemption and how all of the work we've attained to this point could be undone if these big tech companies follow the same pattern that we've seen historically from the tobacco industry.
Dave Bittner: Well, and one of the interesting things this article that you're pointing to -- this editorial points out is that Philip Morris and the tobacco companies did not succeed in passing a federal law with preemption.
Ben Yelin: God knows they tried.
Dave Bittner: Yeah, it's -- and this article says Congress held strong.
Ben Yelin: Congress did hold strong. I mean, I think the movement got stronger. There was more political will around restrictions on smoking, and I think that continues to this day. Even a bipartisan Congress with a Republican Senate and a Democratic House a few years ago passed legislation raising the federal age to -- or the federal floor to buy cigarettes from 18 to 21, and this was supported even by members from tobacco-loving states. There just has been a sea change on this issue. So at some point, maybe it's the public will that supersedes everything. If people become angry enough with the big tech companies, you'll get the ideal legislation in the eyes of the ACLU and others that sets that federal floor and allows for states to be creative in addressing solutions. The big tech companies will be unhappy, but at that point, it might be okay politically for members of Congress to go against the big tech companies, if the public urgency gets that high.
Dave Bittner: I remember back in the mid-'90s, my wife and I were traveling, and we found ourselves in Winston-Salem, North Carolina, and we went out to a --
Ben Yelin: Well, tobacco country.
Dave Bittner: Well, we went out to dinner one night, and the hostess at the restaurant greeted us, and I said we'd like a table for two in the non-smoking section, and she paused and looked and smiled and said, "Bless your heart, you're not from around here, are you?"
Ben Yelin: Yeah. When you literally have a cigarette company named after your town --
Dave Bittner: Right.
Ben Yelin: -- that's going to be a tough sell.
Dave Bittner: Yeah.
Ben Yelin: I mean, look, I think we're not appropriately thankful -- those of us who care about public health and the dangers of secondhand smoke -- how far the movement has come. The fact that my kids don't really recognize the smell of cigarettes. When they're exposed to it, they're like, What is that?
Dave Bittner: Right. What is that smell? It doesn't -- it's not quite weed.
Ben Yelin: Exactly. Now, they certainly know the smell of weed --
Dave Bittner: Right.
Ben Yelin: -- because we're in a state that's legalized it --
Dave Bittner: Right, right.
Ben Yelin: -- but they don't recognize the smell of cigarette smoke, which for me was just like always in the background, no matter where I was. And I grew up in San Francisco, and you'd think that would be the place that would be most resistant to cigarette smoking, but it wasn't.
Dave Bittner: Yeah.
Ben Yelin: So we might just end up seeing the same type of sea change. I think it's kind of a race against the clock. Big tech wants to lock in a federal floor with preemption to prevent states from going further before the backlash becomes too severe.
Dave Bittner: Right.
Ben Yelin: And if members of Congress play the odds and think that they can come up with more robust protections without federal preemption, once big tech companies get even more popular, then maybe waiting it out is in their best interest.
Dave Bittner: Yeah, that's interesting. All right, yeah, it's an interesting analogy, and so we will have a link to that op-ed in the show notes here. My story this week is about the grocery store chain Kroger's. Are you at all familiar with Kroger's?
Ben Yelin: Oh yeah, I don't think I've spent much time at Kroger's, but I've certainly heard of it. I've been to a Kroger's.
Dave Bittner: I don't think I've ever been to a Kroger's. I don't think there are Krogers around where I live.
Ben Yelin: Well, maybe after the segment they won't let you in anyway. So --
Dave Bittner: Well --
Ben Yelin: -- it might not be a problem.
Dave Bittner: -- you know what? That would make total sense because this story is about Kroger's using facial recognition, and so in this case they could see me coming, literally, and lock the door and keep me out.
Ben Yelin: There you go. Uh-oh, it's Bittner again.
Dave Bittner: Right.
Ben Yelin: Institute full lockdown procedures.
Dave Bittner: Right, a big cage drops out of the ceiling --
Ben Yelin: You know, there's a trap -- there's a trap --
Dave Bittner: -- and it falls around --
Ben Yelin: -- there's a trap door below the entrance.
Dave Bittner: That's right.
Ben Yelin: You fall into the fire pits.
Dave Bittner: So this story comes from the folks over at the Record, which is the news department at Recorded Future, and it's about Kroger's plans to implement facial recognition. And evidently, they are the largest grocery chain in the U.S. So they are planning to use facial recognition to target advertising and data collection on their customers, and this has gained the attention of a handful of lawmakers. Congresswoman Rashida Tlaib --
Ben Yelin: Talib, yup.
Dave Bittner: -- Tlaib? Sorry, yeah, it's a hard one to suss out. Elizabeth Warren, who I would say is a usual suspect when it comes to these things, and Bob Casey, a Democrat from Pennsylvania, they're concerned about pricing and privacy risks. And of course, anytime you have facial recognition in any kind of public sphere, we talk about racial bias.
Ben Yelin: Totally.
Dave Bittner: And so there's a concern here about that.
Ben Yelin: And the evidence has gotten pretty overwhelming that facial recognition software is biased against Black and brown communities. I think we're past the point where there's much question as to whether that's the case.
Dave Bittner: Right, right. One of the big fears, and I suppose maybe, well, I don't know if it's the biggest fear when you compare it to the racial bias concerns, but maybe the most concrete fear, the easiest one to realize is a practical issue, is dynamic pricing, which is this idea that the Kroger chain sees you shopping. And by using electronic price tags on the shelves at the grocery store, as you wander around and look at the things you are considering buying, it's plausible that they could change the prices of things based on who you are.
Ben Yelin: So basically, it would be like cookies, but in real life where when -- if you go to Expedia and search for airfares --
Dave Bittner: Right.
Ben Yelin: -- the next time you go, they realize you're desperate, so they can show you the highest prices --
Dave Bittner: Yep.
Ben Yelin: -- because you've left those little -- nice little cookie crumbles.
Dave Bittner: Right. Well, and Amazon famously experimented with this. I don't know if they're still actively doing it, but there have been plenty of documented cases where, if you and I went and looked up the exact same thing at Amazon, we could get different prices because of where we live and all the data that Amazon has on what we do for a living and, you know, what our perceived ability to pay is.
Ben Yelin: Right, right.
Dave Bittner: So that's a concern as well. And then of course there's the data security issues about, you know, if Kroger's collecting all this information, how are they going to protect it, and why collect this information at all?
Ben Yelin: And are they going to sell it? And then it's not just Kroger who has a dossier on you about the items that you buy. Maybe it's --
Dave Bittner: Right.
Ben Yelin: -- other supermarkets.
Dave Bittner: Right.
Ben Yelin: Maybe they sell it to local law enforcement officials. Like we've noticed some odd behavior of people looking at the security cameras very carefully and trying to find the nearest exits. Maybe this person presents a armed robbery concern. Here you go, federal -- here you go, local law enforcement agency.
Dave Bittner: Yeah.
Ben Yelin: Here are some faces for you.
Dave Bittner: Yeah, I mean it's interesting, the sale of the data because, you know, we've already got that with the grocery store customer loyalty programs, you know, where you put in your phone number in exchange for getting a discount on things.
Ben Yelin: Yeah, I was just going to say that. We already have that, but we don't have -- what we don't have is that level of dynamic pricing. No matter what the grocery store knows about me -- and they know a lot about me since every time I make a purchase, I give them my phone number --
Dave Bittner: Right.
Ben Yelin: -- they're not -- they're not changing the prices on the fly when I walk by, right? The prices are the same for every single person who walks into the store. I think the concern is maybe they would not put prices on items. It would say like price will be revealed at cash register. You go to the cash register, and they raise the price on you because they have data that you're desperate to purchase that item.
Dave Bittner: Right.
Ben Yelin: Yeah.
Dave Bittner: Right. I -- yeah, I also wonder, let's say, the system is tracking you as you walk through the store, and you're sitting there in front of an aisle of breakfast cereals, right? And you're looking at, you know, you're trying to decide between Lucky Charms and Captain Crunch -- they could track that.
Ben Yelin: Right.
Dave Bittner: They say this consumer spent this amount of time looking at Lucky Charms and Captain Crunch and, you know, picked up the box and, you know, compared the prices and ultimately selected this one.
Ben Yelin: Right.
Dave Bittner: And that has value, and, you know, whose business is that?
Ben Yelin: That's actually very valuable market research that these companies could purchase and say, "All right, well, what is it about our display? What is it about our placement in the store that got this person to buy one cereal box over another? I guess even if you don't think that Kroger nor any person who bought the data would use it for nefarious purposes, it does give them a lot of incentive to collect information through facial recognition because it's very valuable to them.
Dave Bittner: So Kroger denies that they're going to do any of this, or certainly they deny the use of surge pricing. They're not denying that they're going to monetize the data because, you know, they already do that with the loyalty programs.
Ben Yelin: Right.
Dave Bittner: They say that their focus is on lowering prices for customers. I guess now's the time to note that the Federal Trade Commission is attempting to block Kroger's acquisition of the Albertsons grocery store chain, saying that it would create too large a monopoly on -- or citing anti-competitive concerns.
Ben Yelin: Yeah, I mean that kind of hangs over all of this. It's not a great look for Kroger in the middle of this FTC investigation of whether to block Kroger's purchase of Albertsons. There's a political element to it here. I mean, this FTC has been particularly aggressive on anti-competitive practices and monopolies. Maybe Kroger just waits out this FTC and gets a more friendly one in the next presidential administration.
Dave Bittner: Yeah.
Ben Yelin: But this is certainly bad PR for the time being. You probably want to stay on the Fed's good side if you're trying to get them to permit you to make this giant purchase of another grocery chain.
Dave Bittner: I guess, you know, looking at the big picture here, though, the fundamental right to not be included in a facial recognition program currently does not exist.
Ben Yelin: Right.
Dave Bittner: Even in places like California, where, you know, they have privacy laws or more robust privacy laws than much of the nation, am I correct there that -- is -- do you know of any states that have an ability to easily opt out of facial recognition?
Ben Yelin: No, I do not.
Dave Bittner: Yeah, I don't either.
Ben Yelin: I mean, I don't think that exists. I mean, there are restrictions on the use of facial recognition technology, but there are -- and you could always just not go outside. I don't think they could surveil your -- you inside your house and use that for facial recognition purposes.
Dave Bittner: Yeah.
Ben Yelin: But there are no -- that I know of -- there are no restrictions on places of public accommodations --
Dave Bittner: Right.
Ben Yelin: -- using facial recognition.
Dave Bittner: Well, it's the old thing there's no expectation of privacy in a public place.
Ben Yelin: Right.
Dave Bittner: Which is --
Ben Yelin: And people know or should know that there are cameras around --
Dave Bittner: Yeah.
Ben Yelin: -- even at Kroger's.
Dave Bittner: Right, but I guess this is one of those examples of technology potentially outstripping the practicality of the existing legislation, right? You know, no expectation of privacy in a public place is -- has a different meaning when there are cameras everywhere that can actually recognize you and then track you from place to place just because, right?
Ben Yelin: It's completely different.
Dave Bittner: Right.
Ben Yelin: And the law doesn't really recognize that yet. We're working on it. I mean, not just me and us fellow legal academics, but I think judges and justices are wrestling with this question if we need to change our outlook on the expectation of privacy, given the change in this technology. And different cases have gone in different directions on that issue, depending on the circumstances. But I guarantee you, you'd get a very interesting case if somebody was arrested based on data collected from the Kroger's facial recognition system. I can't really imagine a great hypothetical because, obviously, grocery stores have surveillance cameras, so it would have to specifically relate to the facial recognition element of it.
Dave Bittner: Right.
Ben Yelin: But it's certainly something I could see happening in the future.
Dave Bittner: Yeah. All right. Well, I mean, we'll keep an eye on it here. It's an interesting development, you know? It just feels yucky to me, right?
Ben Yelin: Yeah, it really does, yeah.
Dave Bittner: Yeah.
Ben Yelin: I mean, I get they're trying to get a competitive advantage, but you wonder if it's really worth it, given the PR hit they're taking here.
Dave Bittner: Right.
Ben Yelin: I'm sure in 10 years everybody's going to be doing it, so it's like, oh, maybe Kroger was wise to get ahead of the curve.
Dave Bittner: Right, just chipping away, you know --
Ben Yelin: Yeah.
Dave Bittner: -- they're chipping away, just boiling the frog.
Ben Yelin: I was just going to say that. We are the frog in the boiling water.
Dave Bittner: All right, well, we will have links to all of our stories in the show notes, and again, we would love to hear from you. If there's something you'd like us to cover on the show, you can email us. It's caveat@n2k.com. [ Music ] Ben, you recently had the pleasure of speaking with Eoin Hinchy, who is the co-founder and CEO of a company called Tines, and you all were talking about distraddling EU and U.S. regulations, and of course touching on AI as well. Here's Ben's conversation with Eoin Hinchy.
Ben Yelin: So today we're going to be talking about the regulation of artificial intelligence, and I wanted to see if we could start at a really high level. You can give us the lay of the land. Can you start by talking about the bill that Governor Newsom just vetoed in California?
Eoin Hinchy: Sure, absolutely. So the bill that was just vetoed was one of the -- oddly one of the more popular bills that we've seen bipartisan support for, for a while, and essentially it was set out to allow regulation of large AI providers, and it was generally quite popular. Some polls had 80% of people supporting it. Others had a little bit less, like 60%. What was really interesting was it was essentially designed to provide transparency around how and when AI was being used. And you're absolutely right. It was just vetoed by Newsom a couple of weeks ago, and it's gone back. The main reason it was vetoed was because Newsom said that it was more indexed and biased towards large model providers rather than some of the smaller companies.
Ben Yelin: And how does that contrast with what you've seen in the European Union? I mean, was the California proposal similar? Has the EU tried to go further? How do how do you see the comparison between the two?
Eoin Hinchy: Yeah, so the EU AI Act, which is already coming into force and had actually been around for a couple of years, and you can think about the EU AI Act, or at least the way I think about it is it's less like some of the other recent technology legals that we've seen specifically GDPR, which was around fundamental rights, and the EU AI Act was more around system and product safety. So think about more along the lines of how AI is going to be used in critical systems like lifts and medical devices and so on. And whereas the Newsom bill, which was rejected was more around like general AI and LLM model creation and how they were going to be used. So I guess the most fundamental difference is that the EU AI Act was passed and ratified and is now being put into law. That's the most fundamental thing. But I think the second thing is that the EU AI Act was much more comprehensive in terms of how it thought about AI systems and how those AI systems were used. And an interesting fact about it, I think, which grounds everything, is that it had already been in discussion before LLMs and GenAI began to enter the public consciousness, you know, about two years ago. So this was already in discussion in like 2019, 2020, and so on. And the EU AI Act is it's worth taking a couple of steps back because there is some interesting, peculiar misdirecting language in the whole thing. And essentially, what it does is it sets out to categorize different types of risk and systems of risk where AI will be used, everything from unacceptable risk right through to kind of like low risk, and the various guardrails that should be applied across those -- across those different categories.
Ben Yelin: So it seems like a lot of our AI regulation has been kind of segmented or siloed. I know Governor Newsom vetoed this broader AI policy bill, but for example, he signed a bill on the use of synthetic media in elections. It seems like the EU Act is focused on uses. Do you think that we can kind of put a patchwork together in the way that we've been doing it, or do you think we need something more comprehensive, kind of an all-inclusive approach that does away with this sort of piece-by-piece regulatory framework?
Eoin Hinchy: I think -- it's a great question, first and foremost. I'm a little bit biased because I'm a startup founder, and I'm a big believer in incrementalism and starting small, getting something working, and iterating on it over time. And so, what I would have loved to have seen would have been the acceptance of this Newsom bill and then like improvements on it over time. So sure, let's maybe just start with the large models and the large providers, and then we can iterate and improve on it over time. Because as you know, this technology is advancing so quickly, and my big fear with any technology regulation is that we'll try to do something large, but the technology will have moved on so quickly. But by the time it becomes law or is signed off, that'll no longer be relevant to the actual real-world use cases.
Ben Yelin: And how do you craft regulations with that in mind? I mean, I'm thinking specifics here. Like the difference between ChatGPT-3, for example, and ChatGPT-4o is just monumental. I mean, it's beyond just like a generational update. So how do you craft regulations that can be flexible enough to deal with these just rapid advances?
Eoin Hinchy: It's so hard, and it's the -- it's the constant balancing act, as you know, between like regulation and technology. And I think -- I think the regulation needs to be broad enough to encompass the main concerns and the main kind of like priorities, and then it can be iterated on, and you can have amendments to deal with the kind of like edge cases that inevitably come up. So I think what the EU has done quite well in the EU Act is kind of iterate and learn from that and do the big things. So do it like, look, we're in in life-affecting situations. We're going to have these sets of rules, and sure, there's going to be a lot of specifics and holes that are -- need to be patched up, but at least it's something, and it allows us kind of like lay a foundation that we can then build upon.
Ben Yelin: Do you think, I mean, just somebody who's been in a startup world, that you worry about this regulatory capture where the innovation that we've seen in the industry over the past few years starts to get suffocated just because you have companies dealing with these differing regulatory regimes, and how do you kind of look at that -- at that problem?
Eoin Hinchy: I -- in general, I'm a fan of regulation in these kind of like broad life-affecting technologies. I think it's important, but at the same time, this is the last thing I want my product engineering and legal teams working on. I don't want them having to like spend hours upon hours like understanding or like understanding the vagrancies between these different jurisdictions and so on. But I also think that there's ways companies can build technology that accounts for these types of at least common-sense regulation. And so, what we have done with Tines is, when we were implementing AI, as an example, we built it in a way that was extremely transparent and built it in a way that would allow our customers know when AI was being applied, understand how and when they could opt out of AI, and ultimately giving them the freedom to choose how and when AI was being used in a very transparent way. And we did this like, you know, months and months and months ago. And I think this is essentially what the regulation has recommended, like in in broad maths, is look we want to make it very clear when people are using AI. We want to make it easy for people to opt out if they choose to do so. And so if companies are investing the time and energy in doing this in a way at least the way their customers want, a lot of times the gulf then between what you've already built and what the regulations are stating is not that much work to adhere to.
Ben Yelin: And where do you see this going over the next several years? I mean, obviously here in the United States, it's hard to get anything done at the federal level. That would be ideal because, you know, dealing with a 50-state patchwork is difficult, but like what do you see as kind of the sentiment out there? Are we going to see more stifling regulation? What's the kind of political environment that you're seeing?
Eoin Hinchy: In this case, in the AI case, I'm seeing broad support on Main Street for regulation, and that isn't always the case, right? Like a lot of technology stuff, where like people are just like, No, no, let the market sort it out. And -- but in general, I think for AI, we're seeing a lot of support on Main Street, again, for regulation, and people don't trust big technology companies to self-regulate in this space. And so I think we will inevitably see regulation come in at a state level within certain places. I think California will probably be one of the first states to do it.
Ben Yelin: Always is, yep.
Eoin Hinchy: And then I think -- exactly. And then I think we'll see general acceptance across like -- or not general acceptance, but like broadening acceptance. I also think that what we'll end up seeing is because the EU AI Act is being built out and going to be applicable in the EU, a lot of companies are going to use that. At least a lot of big technology will use that as a baseline, and so we saw this with GDPR.
Ben Yelin: Right, yep.
Eoin Hinchy: Companies weren't willing to build a different product for the EU as they were for like other markets, and so they'll just go look -- you know what? -- we're just going to build this to adhere to EU AI Act, and we'll just roll that out globally because it'll be less of a -- of a carry cost long-term. And so I think that by the time then regulation, whether it's federal or state level is rolled out, I think a lot of technology companies will have already done the work necessary to adhere to the eventual regulations.
Ben Yelin: The other thing I wanted to get into is kind of the psyche around these big tech companies. I think they are largely distrusted in a lot of circumstances, based on our past experiences, and also, you know, they have some CEOs that rub people the wrong way, which I understand. How do you think regulators deal with those unique challenges? Are they -- and do you think the -- kind of the big tech CEOs out there are willing to work constructively on this from what you've seen? What is that relationship like?
Eoin Hinchy: I think they are. So broadly speaking, I think big tech CEO companies do want to build technology that's altruistic. They want to build technology that's like broadly -- sure it's adding shareholder value, but that does good as well. Like I don't think there's many CEOs out there who want to build a product that like that causes widespread harm. That's just like not the reason people start companies, or at least run companies. And so I think there is -- I think there is broad support for some common-sense regulation. I think where the friction comes in how -- is how that regulation comes about. And the bureaucratic nature of a lot of these things just grates and cause frictions with the way a lot of big technology companies like to operate and like to grow. They like to move really fast and build stuff really, really quickly, and that's just not the way regulators operate. That's not how they're wired, and so I think it's important that these -- there needs to be some sort of compromise on both sides, that the regulators are willing to move a little bit faster and be more responsive, and that big tech companies and CEOs and policymakers within tech companies are willing to bend a little bit in order to work with regulators.
Ben Yelin: Some of this -- and maybe this is an unfair question for you, but some of this is just failure of imagination about consequences. So I'm wondering like if we can put on our futurism hats here, five, 10 years down the line, what are some of the AI uses that concern you, that make you think that, wow, we really need to get ahead of this, or we can face some serious kinetic danger?
Eoin Hinchy: Oh, wow, that's such a great question. So the way -- the way I think about it is one of the most high-risk use cases, for want of a better term, where not having humans involved could cause all kind of catastrophic damage or, you know, systemic damage, and so for those kind of use cases for me are things like medical procedures, airline travel, car travel, like all those types of areas, and they're already -- if you think about it -- they're the areas that are already heavily regulated. And so, for me, a really kind of like natural place to start and have concerns is what are the areas that are currently most regulated? Like is it, you know, it's airline travel. It's building. It's healthcare. It's all these areas. And you know, they're also the areas that like are set to benefit a lot from AI and autonomous technologies. But I also think we kind of need to start with the areas that we've spent decades working with regulators to make safe and ultimately provide benefit for the human race and start there.
Ben Yelin: So what do you think companies can do to prepare for regulation, whether it's from the European Union or elsewhere? How should companies be approaching this going forward?
Eoin Hinchy: And so the way we've been thinking about it is we should continue to build the type of AI features that will build trust with our customers and start there, and listen to what our customers are requesting and asking, and giving them the options and the freedom to choose how AI is implemented in their usage of the technology. [ Music ]
Dave Bittner: Interesting stuff, Ben, particularly talking about, you know, potential AI regulation and safeguarding against AI models and so on and so forth. What was your takeaway from this conversation?
Ben Yelin: Yeah, I mean, for one, it goes hand in hand with our first story on this episode because certainly preemption is an issue when we talk about AI regulation, just as it is when we talk about data privacy regulation. A couple of things stood out to me. I think the European Union approach, which focuses on use cases of artificial intelligence, things that present a danger when you're talking about controlling large vehicles like airplanes and trucks and all of that, or critical infrastructure, for example, that's when the standards get more stringent. One of the problems with the California bill, which was vetoed by Governor Newsom -- and we talked about that in the interview -- is the regulation was just a little bit over broad, and it wasn't narrowly tailored to foster the development of greater AI tools while also protecting the consumer. So I thought it was a really interesting conversation. I hope you all enjoyed it.
Dave Bittner: Yeah. Well, again, our thanks to Eoin Hinchy for joining us. Again, his company is called Tines, and we do appreciate him taking the time. [ Music ] 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. We're privileged that N2K CyberWire is part of the daily routine of the most influential leaders and operators in the public and private sector, from the Fortune 500 to many of the world's preeminent intelligence and law enforcement agencies. N2K makes it easy for companies to optimize your biggest investment: your people. We make you smarter about your teams while making your teams smarter. Learn how at N2K.com. This episode is produced by Liz Stokes. Our executive producer is Jennifer Eiben. The show is mixed by Tré Hester. Our executive editor is Brandon Karpf. Peter Kilpe is our publisher. I'm Dave Bittner.
Ben Yelin: And I'm Ben Yelin.
Dave Bittner: Thanks for listening.

