Changes to US AI policy landscape.

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 Cyber Health and Hazard Strategies. Hey there, Ben.

Ben Yelin: Hello, Dave.

Dave Bittner: On today's show, Ben and I are once again joined by our N2K CyberWire colleague, the author of the "Caveat" newsletter, Ethan Cook. Ethan, welcome.

Ethan Cook: Good to be back, guys.

Dave Bittner: Today we are discussing the recently signed White House executive order governing AI. While this show covers legal topics and Ben is a lawyer, the views expressed do not constitute legal advice. For official legal advice on any of the topics we cover, please contact your attorney. All right, let's jump into things here. And Ethan, we're going to trust you to lead the way here. Why don't you set the table for us? We just recently had this executive order from the White House. Fair to say it was long anticipated.

Ethan Cook: Yeah, this has kind of been a several weeks long saga where the Trump administration, there's been some rumors and reports coming out that the administration was preparing to sign a pretty hefty AI executive order. Now, what we got was not what was anticipated because the original version of this order was supposed to be a more encompassing, more restrictive AI-focused executive order that would have established a federal regulatory oversight process for models that were going to be released to the public, yada, yada, yada. After iterations and some pushback from, I think, the people in his camp that are very pro deregulation, we got a much more watered down version of this executive order.

Ben Yelin: I hate to interrupt you, Ethan, but just like --

Ethan Cook: No, go for it.

Ben Yelin: -- to emphasize how close this was, the original one was to being released. Like they had no event planned to release it.

Ethan Cook: They had a whole date picked out. They were ready to sign and they canceled it within a week, like within a week of that sign date. It was, it was a, this was going to be a major turning point, I think is the best way to describe it for this administration on how they've approached AI. And for context, for those who have not been following the playbook, this administration has been very anti-AI regulation at all. They have actively, in fact, gone after AI regulation at the state level. They've tried to pass moratoriums, banning state AI policies, as well as have established federal groups to sue and file lawsuits to shut down state AI policy. And I think the real turn, which if you hear all that and you go, well, why would we want to regulate it now? Like, what was the pivot? I think the simple answer to that is Mythos. Cloud Mythos single handedly cause them to realign their AI policy stance.

Dave Bittner: Well, let's see, why don't we start off with what is in the one that was actually signed and released and sort of go over the breadth and depth of what it outlines. Ben, I know you've studied this as well, so I'll let the two of you fill us in on what we're talking about here.

Ben Yelin: So, yeah, I mean, I guess I can get started and talk a little bit about this compromise that they reached at this high level meeting. Which was attended by cabinet secretaries, but also the main AI advisor who's now outside the administration, David Sacks, who I think played a big role in this. So the way this executive order works is that the government would get an early look at powerful new artificial intelligence models, which would give federal officials a chance to brace the economy and the private sector for potential security risks. In the previous iteration of the executive order, the time frame for review at the federal government would be 90 days. And basically, the industry representative said that that was unacceptable because they can't wait 90 days.

Dave Bittner: A whole quarter.

Ben Yelin: Exactly. Things are moving too quickly. You're going to be suppressing our chance to get to the market on this and get ahead of our competitors. And if you want people to participate in this program, and again, it's voluntary, this is not something that's mandatory, then we're going to have to shorten that review point. So the date they settled on, or the number of days they settled on here is 30 days. Which is something that David Sacks, I know, has been pushing for. It's going to mean that the timeline for reviewing these new models is going to be extremely fast, and obviously that can have downstream consequences. They might not have time to find the type of risks or security flaws they may have been able to find in a 90-day review period, but that's the compromise that they reached here.

Ethan Cook: Yeah, and I think there is, you know, the one of the other things they've asked, part of this, you know, 30-day process is to establish a, a clearinghouse of sort, I think is the way they described it, which is going to go under the Treasury Secretary. And that's going to review any of the discovered AI, any of the discovered security vulnerabilities, et cetera. And it tasks various members and various departments under the federal government to start, you know, basically looking at AI from a security perspective. Which I thought, you know, a little bit formative in terms of its language being like, oh yeah, make sure the AI is secure. You know, we're looking at it from a security focus, et cetera. But I think the way the administration is characterizing this is as a common sense approach where the first iteration was far too regulatory in its nature, far too unrealistic, et cetera. But we can't do nothing. So this is a nice, happy medium ground between the two options.

Ben Yelin: And I think the administration, as you said, Ethan, was legitimately moved by what happened with Mythos. I think that was the game changer. And I think it's possible that if we had not had the rollout of Mythos over the past couple of months, the administration would have just punted this and said, yeah, you know, our posture is innovation at the expense of regulation if need be. And our priority is outcompeting China and our other adversaries. And so we're not going to bog, we're not going to bog ourselves down with even voluntary review periods for new models. But then seeing what Mythos was capable of, I think not just the administration, but the industry was like, okay, you need to pay attention to this. And we as the titans in the industry are telling you how dangerous this is. And they have a very influential voice in the administration. I think that was the motivating factor that pushed this across the finish line.

Dave Bittner: What do you all make of a lot of the speculation or I guess criticism is probably a better word for it that I've seen plenty of cybersecurity professionals out there, experts in AI, who have been saying that Mythos is basically marketing, that it's really not as powerful as Anthropic would have us believe. Other models are capable of doing what it's capable of and that this is a successful positioning to make something sound exclusive and mysterious and powerful, perhaps more so than it actually deserves.

Ethan Cook: Yeah, that's a good one. I think there is some credence to say that every time a new model comes out, there is always this very convenient like two to three-week period about the hype of what this model could do. Sometimes it's about promise and it's like, oh my God, look how, we're going to be able to do all these things and it's going to be great. You should really get on this model. And other times it's fear being like, we need to get ahead of this model, so we should really focus on it. There's some, there's some trackable points there that has kind of built over this past couple years that this is a reoccurring factor. So I could see that. I would be skeptical of just automatically casting this as all marketing, I think for a couple of reasons. One, the Pentagon is actively incorporating Mythos into all of its systems.

Dave Bittner: Even though they're banned from using it.

Ben Yelin: Pretty important point to include in there.

Ethan Cook: Exactly. And I think that the fact that they're banned and also have signed with OpenAI and are still using Mythos, I think says a lot about Mythos and what it can do.

Ben Yelin: Says a lot about the government.

Ethan Cook: Exactly. I mean, you know, whether you like him or not, Trump is known to be very antagonistic to people he does not consider to be his friends and allies.

Dave Bittner: No. Really?

Ben Yelin: I'm sorry.

Ethan Cook: And I think it's fair to say that for him to, you know, either turn a blind eye or prove us or the US government by that, I mean, to use Mythos despite having a blacklisting label on them openly, it says a lot about what Mythos can do and whether or not Mythos is everything that it's promised to be. I think there is a general sense that AI models in general and in cybersecurity speaking premises are about every three to five months kind of dramatically improving by benchmark standards, right. And I think there's this dynamic of while Mythos may not be exclusive, it may not be the best in capabilities or may not, you know, be the only model capable of that. It's certainly advanced, and it could be a starting point in the next couple of months. And we're about to hit that three to five month mark since it was first, you know, demoed where we could be seeing a another really more advanced version of this model coming out in the coming weeks or coming months, right. Whether it's by Anthropic or another company who's used its tech or evolved its model based off what has already happened. And I think that's something that is really important to get ahead of. And I think that's something what this, this order is focused on.

Ben Yelin: Yeah, I mean, I think that's the critical point you made there is that, yes, it's about Mythos itself, but it's more it's also what it represents. Which is like with every new iteration of AI, the cybersecurity risk seems to be going up exponentially. And so if you don't institute some type of voluntary review process now, we're going to get to a point in whether it's six months or a year where the models and the release of the models are outpacing our ability to confront these risk factors. And I think that's what they're trying to get ahead of.

Dave Bittner: The fact that this is voluntary, how much does that affect things? Is it, you know, mandatory volunteerism? Is this, you know, in other words, if you want to have favor with this administration, you're going to participate, right.

Ethan Cook: Yeah, I think that's a great question. I think, and obviously this is me speculating, I think for companies that are very interested in working with the federal government or already are, such as like OpenAI, for example, Meta, Microsoft, I think there is a dynamic where they will view this as not voluntary because I think it's a, you know, you scratch my back, I scratch your back kind of thing. We don't, the government's, I don't want to be blindsided. I don't want to not see this coming again as they kind of were caught with their pants down with Mythos. But at the same time, the companies look at this saying, hey, we get in your good graces, we comply, we get nice PR wins, we can potentially avoid any major vulnerabilities, and we don't really have to bend over backwards for this. I think the more interesting question will be what does it look like for the smaller, more niche AI firms that we partner with, not just right now, but over the next year or two, and how does that play out and what is the advantages? And I don't really know the answer to that one. I think that will be a kind of a wait and see situation.

Ben Yelin: Yeah, I mean, I agree. I think the big companies, it's still bizarre to me. They've shown not only a willingness to be regulated, but almost like they are begging the government to put guardrails on them in a way I've never seen any other industry act. And I don't know exactly what that's about. I mean, I think some of them have bought so far into their own hype that they are scared of the impact that they can have. And so they want buy-in from federal regulators I think is largely what it is that.

Dave Bittner: I've seen some people say that this is, the call for regulation is kind of a oh please don't throw me in the briar patch kind of thing of regulation would help keep other players out of the game because the regulation would make it so only the big companies could play here.

Ben Yelin: That's the cynical view, and I think --

Dave Bittner: You don't think it's earned?

Ben Yelin: No, I think it's very possibly earned. You know big companies do that all the time.

Dave Bittner: Right.

Ben Yelin: They're the ones who are going to the like environmental review meetings of big infrastructure projects in California, right, and saying like this is going to hurt the endangered hummingbird. When really they're just trying to stop their competitors from being able to fulfill a contract.

Ethan Cook: With the hummingbirds.

Ben Yelin: Right. Yeah. So no, I would not put that past them. But part of it, I at least think is genuine that, you know, I think they're, they've developed a fear of their own product and I think see themselves as having a place at the at this kind of crossroads of history where they don't want to be seen as the group of five people that brought down civilization.

Dave Bittner: Right.

Ethan Cook: Oh, I also think there's a, not just the leaders, but the engineers and developers who work within the company. Like there are many cases where people have resigned and expressed concerns about this. And there are many cases of petitions from employees, you know, sending to upper management. So I think there's also a degree of I don't want to use the word placating, but like appeasing their own employees and saying like, hey, we are concerned about this too. Like this is something that we're not just blindly hoping for. And so we're calling for regulation. So don't, like don't, there's no need to be as mass panicked from that your job is potentially going to cause the end of civilization for I think there's a perspective there as well.

Ben Yelin: Yeah, like you can still have a good conscience and work for us because we are actively trying to prevent the worst harms from coming to fruition.

Dave Bittner: Right. We've all seen The Terminator.

Ethan Cook: Yeah.

Ben Yelin: Yes.

Ethan Cook: I think an interesting angle to this is that, Ben, you touched upon it briefly and I would be curious to get all your thoughts on it with the China aspect and the, something that is really kind of interesting to me about this is the US has been opening up its AI policy. Trump has allowed the sale of some of Nvidia's still powerful but not most advanced chips to China in limited controlled amounts. There is obviously a very present air of deregulation throughout the country, and they're trying to bring that to the state level. And I think that is this kind of tone of leadership that the US is taking right now. Whereas China is kind of taking the other stance. China is saying that we are trying to internalize. I mean, they're shutting down or attempting to unwind, r I don't really know what the status of it is, Meta's acquisition of an AI or advanced technology firm over in China. And they just re-evaluated the laws. I think they're retaking place on July 1st that is going to shut down and reassess any foreign influence or foreign investment into Chinese advanced technology companies. And so they're kind of having this more closed off like, you know, keep everything in house. And I would, you know, just your thoughts on the juxtaposition between those two viewpoints, as well as the kind of concern that is the US being too liberal with its AI stance and kind of being able to allow some of these technologies to get out.

Ben Yelin: I wonder I mean, they just had this summit, this very high profile summit in China. We don't really have much information on what came out of that summit. Like there wasn't any type of broad bilateral agreement that was signed. But I'm wondering what the impact of that conversation was.

Ethan Cook: And the pivot away from the first iteration of this policy came right after that. Like they had this meeting and they were prepping this original order of this executive order to go through. And then very quickly afterwards, there was like within days or like a week afterwards, they pivoted away from it very quickly.

Ben Yelin: I noticed the timing.

Dave Bittner: Yeah.

Ben Yelin: You know, I don't have enough information to say definitively whether one thing led to another, but the timing, I'll just say it's interesting.

Dave Bittner: Yeah.

Ben Yelin: Yeah.

Dave Bittner: What about the, I guess, ongoing specter of the federal government precluding the state's ability to regulate these sorts of things? There's been fears of that and so far hasn't really happened. We've still got states doing their own thing. Did this AI executive order move the needle on that in any way?

Ben Yelin: I don't think so. I mean, I think it's still the administration's posture that they want to preempt state level regulation of AI. They've tried to do it in a number of different avenues. They weren't able to include it in their reconciliation bill last year. The executive order that they promulgated in the winter was focused on setting up a new division, basically a new division of the Justice Department to sue states who put out these regulations, that's largely discretionary. And we haven't yet seen them employ, at least to my knowledge, we haven't seen them employ that tool to instigate lawsuits against states who are continuously passing AI regulation. Now, some of them are very sector specific or relate to AI governance within a state where the administration has said, like, we'll take a hands off approach to that. Or if it has to do with like child safety, you know, we're not going to go after those laws. But I think in terms of broader regulations and guardrails, I think they're very committed to this idea pushed by David Sacks, but also the other companies that we need to have a regulatory regime that's consistent across 50 states that's not fragmented and where companies can, once they decide to go to market, can have expectations on what regulations are going to entail. And there are proposals in Congress that are floating around. All of them are bipartisan at comprehensive AI regulation. So you could really get those regulations enacted through a statute. It would stand on much firmer legal ground. And then you could expressly preempt through that statute certain types of state laws on AI regulation. But the stasis in Congress is always the thing that prevents it. And even minor disagreements can sometimes derail these bills, even though there seems to be bipartisan enthusiasm for them.

Ethan Cook: Yeah, I think there is a, you know, I'm kind of waiting for the other shoe to fall with this. I think we saw for the better part of a year that the Trump administration was very adamant that they wanted to deregulate. They were, whether it was through nothing at the federal level kind of signaling that if something was passed in Congress, that they would not sign it if they did not agree with it. You know, there was, if it was over-regulatory or overly burdensome, and then going after state level. And then I think it culminated, as we said earlier, with Mythos. And I think there was, and I think there also, you know, David, as Ben said, David Sacks has left the administration in his role as the AI czar. And I think that he was a very adamant proponent of deregulation. I think was he the one, I think it was him that he was the one who said that we can solve AI's problems with AI, you know, had a very, you know, very, very high on just using AI to solve everything kind of stance. And him leaving Mythos coming out where I think we saw the pendulum start to swing the other way. And as it's kind of centered back here, I think part of this compromise of where we're at right now with this order will be kind of a pause on the moratorium stuff. Because if you're going there and saying, we need to have some sort of regulation, but we're not willing to do it at the federal level, letting the states implement that and still allow some regulation to come through, albeit way more limited. And if and then the Federal Government can always lean on, if there is a problem, we can choose to enforce it through our executive order that we that we signed several months ago, or whenever they decide to go after it. Whether it's successful or not is a different perspective, but I think that is kind of how I see this one playing out, where they passed it, it's there. I don't get the impression that they really intend to use it. The fact that they haven't launched anything yet is a huge driver there. And you know, if that was passed at this point almost, you know, several months ago, and they haven't done anything on it, I don't really see them making waves on that unless they really feel like a state is being overly aggressive. I mean, maybe they go after a California law, that could be it. But at the moment --

Ben Yelin: Gavin Newsom sends a mean tweet.

Ethan Cook: Exactly. But I, you know, especially with how bipartisan some of these efforts are, I mean, to your point, Ben, at the state level, this is going on, too. You have Florida going after OpenAI. I know you guys just talked about that.

Ben Yelin: We just recently talked about that.

Ethan Cook: And, you know, this isn't something that is really a conservative and progressive viewpoint. It's more of a ow is this impacting state by state and what do states feel individually is the best way to approach it is the vibe that I'm getting.

Dave Bittner: We're going to take a quick break to hear from our show sponsors. We'll be right back after this message. [ Music ] If we are at this societal inflection point, and I think many people think that we may be. I think some people think that maybe there's breathless overenthusiasm for that notion. But it seems like people in the administration and in industry are both taking this seriously. And if the leaders of these AI companies are indeed worried about how this may play out, I'm wondering like where that takes us, you know, because, this companies are not as, as everything's being run right now, everybody agrees is not sustainable. The degree to which they are throwing money into a furnace, right, to make these things happen. People are underpaying for the services they're getting. That can't go on forever. And we've even got things like, you know, Bernie Sanders' proposal that these companies be partially nationalized or the federal government invests in them so that the citizens have a stake in the outcome of these sorts of things. I mean, I guess what I'm getting at is like to what degree is this new ground, and we're not sure how any of this could play out.

Ben Yelin: I mean, I'd call it the Bernie Sanders-Donald Trump proposal. They obviously are coming at it from different perspectives, but --

Dave Bittner: Right, strange bedfellows.

Ben Yelin: Right, I mean, I think both of them have an instinct that like, if the federal government is going to be absorbing the risks of the AI revolution in the form of economic displacement, massive changes to the labor market, overuse of critical infrastructure, et cetera, then we should be able to reap some of the rewards as well. And like that's a pretty socialist viewpoint. That's why Bernie Sanders supports it and is putting forward a proposal. But it's also something Trump has talked about in other contexts before, including within Nvidia, where, you know, he thinks that if the government is going to be affected so heavily by the success or failure of these companies, then we should at least have some type of stake. I mean, that's where he comes from in the world of business before he got into politics. So I mean, talk about strange bedfellows.

Dave Bittner: Right.

Ben Yelin: But I find that aspect of it really interesting.

Ethan Cook: I think there is a dynamic where as we continue to, you know, you mentioned how we kind of feel at this like inflection point of society. I'm a little lower. I don't think we're quite there yet. I think it's going to be a couple of years where we truly hit the point where we hit, I think early estimates are for like truly artificial general intelligence was like 2030 is the kind of the benchmark for some of the initial models. I think that's when we're going to really start hitting these inflection points and seeing the impacts on the labor market, the impacts on jobs as we know it, et cetera. We're starting to see the impacts of that, but I don't think we're fully realized yet. And I think that the premises that Bernie put out, taking stake in these companies, et cetera, I think to me, I'm generally behind it because I don't see it as a fixing the problem now. I see it as getting ahead of the problem that is inevitably going to come in five years. And I'm always in favor of that. And I would rather have a government that's more involved in this technology, not necessarily from a regulatory perspective, but from just a, they understand there's no blind sides. They can craft effective policy, not on the fly or in a reactionary stance, but in a proactive stance and saying, okay, we're getting ahead of this, we're getting a handle. And I think it's fascinating that this isn't a, whether, I don't know what the motivations behind it are, but this isn't a polarized issue. This is something that is generally seen as something that's good. I don't know if both sides agree on why it's good. I think some may have more a social impact perspective, some may have a financial impact perspective. But at the end of the day, if we get to the same point, then I'm good with it either way.

Dave Bittner: To what degree are we seeing this issue out on the campaign trail? Ben?

Ben Yelin: It's kind of slowly but surely making its way into the normal course of political advertisements. I mean, you see it in certain districts, especially locally, where you get proposals about a moratorium on the building of AI data centers.

Ethan Cook: Yeah, New York just passed one. New York just passed one. I mean, I've seen it in smaller rural counties in Maryland where, you know, this is something that they see as politically advantageous and also I think genuinely believe in. That we don't know what the impact of these giant data centers are going to have on our infrastructure. Let's get ahead of it and pass that moratorium.

Ben Yelin: I think we're going to reach much more of an inflection point. Maybe this is when the AGI revolution happens. Maybe this is before that happens where AI and its ripple effects on society are going to be the key political issue of our time. And I think it's at least within the realm of possibility that that happens in 2028, that there's enough job displacement or there's enough strain on our resources, or there's been some type of massive AI enabled cybersecurity incident where it gets more into the public consciousness. I think right now, like there are other things for politicians to fall back on that they know are effective in advertisements, but I think we're close to the point where AI, because of how profound an impact it's having on everything, is going to be a divisive political issue. And I don't know exactly what that's going to look like when it happens, but I think that could be as soon as 2028.

Ethan Cook: Yeah, and I think for me, the way I see it manifesting in the election, I think it's going to, in any election going forward, I think the place will be most felt at the federal level will be in the House of Representatives. Where county by county, especially those impacted, will really, really feel it. Same at the state level. I think that's going to be a pretty big way that that kind of galvanizes support. I think the Senate will see it to some degree, depending on the state. Federally, for the general election, the presidential election, I don't know if this is going to be a big issue. And I don't mean that in a sense of the impacts aren't relevant. I mean that in a sense of the, I think there are fears about it, and it depends on how these moratoriums, I think, play out in the feedback and do they stick? Are they well grounded? Is the research backing it? But I get the sense that both administrations, the Biden administration and the Trump administration, were very adamant that AI investment needed to happen. Differed on how much regulatory oversight there had to be, but that investment needed to happen. Biden loved the CHIPS and Science Act to help bolster that backend supply chain. Trump is more a fan of the deregulation approach. I don't see any, at least currently with where we're at, any administration really changing from those stances. I kind of see presidential candidates taking that stance of this is a necessary thing that we have to employ. And I will let people really express those opinions in different avenues at the state level as well as the House of Representative level.

Dave Bittner: Well, let me ask a really basic question that I think could be a campaign issue. Is AI creating more jobs or is it taking away more jobs.

Ethan Cook: Yeah, fascinating.

Ben Yelin: That's a great question.

Ethan Cook: Great question. I don't think we know, at least I haven't seen any real concrete data that suggests one way or the other. I think it's replacing, but replacing differently than previously. Whereas like data center building jobs are skyrocketing. There's a ton of investment there.

Ben Yelin: Yeah, I mean, I was going to say like, if I, and I am no economist, I did minor in economics. I did stay at the Holiday Inn Express last night. I took a class. Yeah. But if I had to guess, I mean, I think like right now it's a net positive for jobs just because of what you said, Ethan, like we are seeing the manufacturing boom that presidential candidates have been promising and outlining plans on for decades. It just happens to be for the construction of data centers. And those jobs are more and more ubiquitous, whereas you're starting to see some evidence, and it's very tentative, about especially entry level positions, white collar positions being eliminated because AI can do their jobs.

Dave Bittner: Tens of thousands of layoffs in tech this year.

Ben Yelin: Oh, sure, sure. So the tech layoffs. So yeah, I mean, I think it's more sector by sector right now. I think, like I said, on net, right at this moment, it's probably positive, but as with any new technology, I mean, I don't think you can understand the full impact until we're talking about a horizon of decades. Going back to the like industrial revolution through the dot com boom, like it just takes a long time for the labor market to fully adjust to the new technology.

Dave Bittner: Yeah, I remember, you know, I did some documentary work years ago looking at the, the development of planned communities back in the 1960s. And one of the things that they believed was going to happen was with the, with computers, people were going to have much more leisure time. They're going to have a four-day work week, and so we're going to have to have all these facilities for people to be able to take up their leisure time because everyone will share the benefits of the productivity that will come from these computers. And that didn't happen.

Ben Yelin: Whoops.

Dave Bittner: Yeah.

Ben Yelin: It's so funny because now we're having the same debate. Like people like Elon Musk are saying, well, most people won't have to work.

Ethan Cook: Right, universal basic income.

Ben Yelin: Yeah, universal basic income. It's delusional because people still need to find a purpose in their lives and they're not just going to sit back and be like, yeah, I'm cool doing nothing and just taking, you know, the median income as a check every month.

Dave Bittner: And let's not forget, I mean, there is a whole, a large, sizable group of people who have a core belief that people should not be paid for doing nothing.

Ben Yelin: Absolutely. If you know one thing about politics, I think that is certainly the case.

Dave Bittner: Yeah.

Ben Yelin: This is going to sound like a super liberal thing to say, but I'm going to go for it. I'm just going to cite there was a very interesting Ezra Klein column on this topic in the New York Times where, I mean, I think the way he sees it is this is not going to displace jobs just like everything else. This is going to be an accelerator. It's just going to change the nature of jobs. You know, you're going to have to have analytical skills. It's just a new sophisticated type of analytical skill that you didn't have in the past.

Dave Bittner: Yeah.

Ethan Cook: Yeah, I think to that point, there is amongst the meta, recent meta layoffs, there was a sizable amount of layoffs, but there was also a sizable amount of job reframing where they were taking thousands of their employees and saying, okay, you used to do X, we're now having you do Y. Whether that's because it's different with AI or because we want you to pivot focus to AI projects. And I think that to your point, that is a, people are very focused, and rightfully so, on the layoffs and the impacts that that could have. But I think there is another under sweeping thing that's not getting as much attention, which is how many people are having their jobs fundamentally change because of AI.

Dave Bittner: I don't know, guys. I mean, I keep thinking about, you know, we're going to retrain coal miners to be developers. Learn to code. Right. Like, come on.

Ethan Cook: No, I agree. I think there's absolutely a factor.

Dave Bittner: I think there's so many, I look back, you know, again, my previous career in the television industry, I remember when robotic cameras came out and you'd see ads in the trade magazines that say, you know, thanks to these robotic studio cameras, you'll be able to reassign your camera people into more meaningful jobs at your organization. No, they're getting fired. They're getting let go, you know. And so I guess I'm, I guess I'm cynical from experience.

Ethan Cook: I think rightfully so. Yeah.

Ben Yelin: There's the macro pointy headed economist view, which is like --

Dave Bittner: Right.

Ben Yelin: all the jobs will be replaced by better jobs. But then there's the like actual human lived experience where I worked at the same frickin factory for 40 years. I could afford a nice house on one income. You know, education, higher education costs significantly less than it does now. I don't care that, you know, macro sense, like my 25-year old kid is eventually going to have a better job than me.

Dave Bittner: Right.

Ben Yelin: Like it's, it's about me right now. Yeah.

Ethan Cook: I think it's the dynamic of societal impact and human societal cost and human cost, right. I think on the macro to your point, Ben, there is a, you could say it is a net positive for society, but that doesn't mean for every human it is net positive, right. Like there are plenty of people who are not going to be able to find jobs that are quote/unquote replaced quickly. Or maybe that in their area, there is no replacement job because the data center hasn't been built in their area to go invest in that. Or the tools haven't been built fast enough. Like if we're still building these programs that are designed to upskill people and get them pivoted, okay, great. But if that takes 10 years to build or five years to build effectively, what is that person supposed to do for five years after they've lost their job? Because they're not losing their jobs in five years. They're losing them, some of them right now.

Dave Bittner: Well, let's not forget, building data centers requires a lot of people. Running data centers does not.

Ethan Cook: Yes.

Ben Yelin: Yeah. And eventually we have to like, eventually there's going to be enough of them built. We're not there yet, but like there's a theoretical end point where --

Ethan Cook: Or shipping them to the moon.

Ben Yelin: Yep, or Mars. I mean, probably more likely it's going to be first China. But yeah.

Dave Bittner: How about the bottom of the ocean?

Ben Yelin: Yeah, I saw that. I saw that.

Dave Bittner: Cool, easy to cool them, you know, water, all that.

Ben Yelin: I've seen some great.

Ethan Cook: My favorite commentary was having someone's like, if we do that, the octopuses are going to become smarter than us. And they're going to take over.

Dave Bittner: There you go, sure.

Ben Yelin: I've seen some great AI generated images of like Central Park becoming a giant data center.

Dave Bittner: Yeah, yeah.

Ben Yelin: It's just our future.

Dave Bittner: All right, let's bring it home, gentlemen. Closing thoughts here on the AI executive order. Overall, let's wrap it up here. Let me start with you, Ethan. What's your takeaway?

Ethan Cook: Yeah, I think I would say that overall, I like it. I think it is a step in the right direction. I don't love it because as someone who thinks AI needs a little bit more of a stronger hand on it, and this isn't a hot take because AI companies have called for this. So this is a reasonable approach, I think. I would like to have seen, if you're not going to do 90 days, at least do 60 or at the minimum of 45. One third of the previous original proposal seems a little weaker than I would have liked it. I think that with Sacks now out of the administration, I'm sure there's some influence still there, but with him out, we may see some more of this coming out a little bit more aggressively over the next, by aggressively, I mean over the next like two years, not in the next three months. And I think that we are one model away from the impact that Mythos had, not from a literal actual capability perspective, but the, the perspective it has or the long-term optic from us having a way more robust framework being put in over the next six months to a year.

Dave Bittner: Ben.

Ben Yelin: I think 30 days is concerning, as you just noted, Ethan. I think the other thing we need to see is the administration is going to be setting classified standards which are ultimately going to be finalized at the administration level. And until we know how robust those standards are, you know, I think we should reserve a little bit of caution as to how much this is really going to change.

Dave Bittner: That is "Caveat" brought to you by N2K CyberWire. We would 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 e-mail to caveat@n2k.com. This episode is produced by Liz Stokes. Our executive producer is Jennifer Eiben. The show is mixed by Tre Hester. Peter Kilpe is our publisher. I'm Dave Bittner.

Ben Yelin: I'm Ben Yelin.

Ethan Cook: And I'm Ethan Cook.

Dave Bittner: Thanks for listening.