The US AI Framework Is a Press Release - DTNS WEEKEND

0:00This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade-in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. Parles-tu français? Hablas espagnol? Parli italiano? If you've used Babbel, you would. Babbel's conversation-based technique teaches you useful words and phrases to get you speaking quickly about the things you actually talk about in the real world.

0:43With lessons handcrafted by over 200 language experts and voiced by real native speakers, Babbel is like having a private tutor in your pocket. Start speaking with Babbel today. Get up to 55 % off your Babbel subscription right now at babbel.com slash acast. Spelled B-A-B-B-E-O dot com slash acast. Rules and restrictions may apply. Taking care of your eyes shouldn't be a hassle. That's why Warby Parker is a one-stop shop for all your vision needs. Our prescription glasses and sunglasses are expertly crafted and unexpectedly affordable. Stop by a nearby store or use our app to virtually try on frames and get personalized recommendations.

1:26Did we mention we offer eye exams and take vision insurance too? For everything you need to see, head to your nearest Warby Parker store or visit warbyparker.com today. That's warbyparker.com. If you work in university maintenance, Grainger considers you an MVP because your playbook ensures your arena is always ready for tip-off and Grainger is your trusted partner. Offering the products you need all in one place from HVAC and plumbing supplies to lighting and more, and all delivered with plenty of time left on the clock, so your team always gets the win. Call 1-800-GRAINGER, visit grainger.com, or just stop by.

2:03Grainger, for the ones who get it done.

2:15This is the Daily Tech News Show Weekend Edition. I'm Tom Merritt. Thank you so much for supporting the show at patreon.com slash DTNS allows us to every once in a while dig into a topic a little deeper. And this week, very happy to have Andy Beach from Engines of Change on the weekend version of the show. You hear him all the time on the regular show. We're going to spend a little more time talking with him about the U.S. AI framework, legislative framework. Andy, thanks for joining us. Hey, Tom. Great to see you again. So the Trump administration released a four page policy framework back on March 20th.

2:52The idea is they want to encourage the acceleration of development. They want to defer copyright questions to the courts. They want to prevent states from experimenting with their own laws on this. And they don't include any kind of funding requirements or provisions for this. There's no enforcement body for this. This is a framework that they are proposing to the U.S. Congress. Here's what we think your laws should look like. These kinds of principles. Andy, what's actually in this suggestion? And keeping in mind that this is a suggestion to Congress and Congress is very favorable to the administration.

3:34But what's in it? Yeah, I should probably I'll set the stage by saying I felt like it was toothless when it came in. And my initial reaction was very, my hackles were up immediately around it, although I settled down a little when I realized it was a framework and not the actual policy itself. If you recall, back in December, the Trump administration had also said that they were planning on a federal AI policy and that states were going to be superseded. Effectively, they were saying states don't have a right to regulate AI. We don't want you in that business. And, you know, please don't waste your time.

4:11Now, we had a couple of states who have done things even in the interim. We had a couple of states that already had something by then. But this is the first time since then. So a full 90 days has passed before we got this framework out there. And I don't love what I see. So what are the things you don't love? Because you've actually read this whole thing, which I commend you for. Oh, it hurts. Yeah. And again, reminding everybody, like the fact that it says we don't think states should be allowed to have their own laws on this doesn't mean that that is the way it is now. This is the administration saying to Congress, we would like you to make a law that says that.

4:50What are the other recommendations in here that you see? So it had a lot of language that was right in terms of it. It covered things like it had information about the data centers in there. It had some child protection language. The biggest issue I think I ultimately have with it is that it didn't support, it did not mandate any kind of funding or direct any given agency to take this on. In other words, there is nobody that can pick this work up and do it, and they don't have any money to do it. So it will the current way that it's written, it would have to become a new task that an existing government entity has to do.

5:33And that's just impractical. Do they prohibit funding of it or is they just kind of leaving Congress to figure out that part of it? Like, could Congress say like, oh, yeah, we'll we'll take this framework, we'll enact it, but we'll add provisions for either maybe even creating an agency, but funding an existing agency? They would absolutely have to in order to make it. But the reality is, to your point, this current legislative body is pretty administration friendly. And they're not necessarily without a mandate to go do something. They're unlikely to just add it on their own. And so I think it just makes it harder.

6:12It means that more people in Congress and in the Senate will have to push to get those pieces in place in order to make it something that is actually enforceable and actually has some some teeth behind it. It is one of the reasons that I feel like this might be a little bit performative. And I'll leave it to Justin Robert Young, who also had a great conversation with you about this, to talk about the more political side of it. But it is not unheard of for administrations to hand over bill text and say, we think it should be something like this. And then Congress obviously revises it. This is not that.

6:49This is just kind of a list of ideas. One of the ideas that I think is important is coming up with new legislation around how we treat copyright with these models, because the existing copyright laws just don't contemplate the idea of using data for training, because it is technically transformative to take your data and turn it into a list of numbers that are weights. The actual information is not copied into the model. However, people seem to not care that that's the case, and they don't like that you can train on your data without your permission. And so that's why I think we need new rules about this.

7:32What is the administration recommending in this framework regarding copyright? Yeah, well, that was another issue that I certainly had with it was that the copyright was sort of a punt. They don't particularly have good language other than the courts will figure it out at the moment. I think if I ran down the list of areas I had that were sort of a bigger problem, the designated oversight body was sort of the big one for me. But the fact that they also had, we don't have a hard deadline around solving copyright, which I think would have also been helpful. Um, we also have, uh, no language in there around allowing the state run, uh, legislation around AI continue until there's federal things in place.

8:23In other words, this framework effectively kills those state legislations, but there's nothing to replace them yet. Yeah. And, and that to me is problematic because to me, the state run legislation will be a great testing ground to see what works, what's practical and whatnot. so that when we get to a federal guideline, it's more mature and evolved. Now we're making the federal guideline have to work harder right from the beginning. And we've got this gap of time where no legislation and no regulation is going to occur. And that's great if you're a technology company because it allows you to sort of rapidly accelerate, but it doesn't give us the guardrails that will keep this from getting out of control.

9:06If that were to be turned into a bill, and if that bill were then to become law. But again, this is just the administration saying, we'd prefer it if you put in your law that states can't do this. And as we know, Congress probably doesn't do most of the things people ask it to. So that is part of my coming to this over and over, saying this is a framework, this is a recommendation. Because even in your response right there, you were saying this would kill. And I'm like, but it would have to actually be a thing for it to do that. Right. That is my lack of faith in the legislative body we currently have in pushing back against these things.

9:47I think one of the things. But again, what would they be pushing back against? There's no bill. There's they have to write a bill first for it to even be something for them to push back against. Sure. But they need a mandate to go write a strong bill. And I don't think they've been given a mandate to write a strong bill. I think they've even been given a mandate to write a bill, though. that's how weak this framework is to me. That is, that is true. Absolutely fair. And I guess for, for a couple of years now, I have, I have looked at what the EU has done around AI regulation and the tech industry has been highly critical of, of the EU and, and have, have lots of VCs have publicly said that it's, it's pushed back and stifled innovation there and allowed it to proliferate in the US.

10:33But at the same time, I think there's a lot of inherent dangers as we're seeing with what AI interactions can be, and there need to be guardrails there. And so I think healthy pushback from a regulation body is actually a requirement for this kind of industry. There's a historical analogy here that we can look at in our very near history, which is the proliferation of social media and the internet and sort of the harm and the addiction that tech companies were able to create around those technologies somewhat accidentally. But at the same time, they also knew what they were doing because they were applying game tactic behavior to it.

11:18And I think if there had been more regulation early on about that, we maybe could have stopped some of that harm, particularly for younger kids that came into it. And this is, to me, we're taking, we're not taking any of those lessons to heart. We're just going ahead with the same playbook we used with the internet a decade or more ago and applying it now to AI. And I worry, what are the compounding effects that we'll see in 10 or 15 years? And so, yes, you're a hundred percent right. They could come back with something that's much stronger. I just worry that what I see right now doesn't lead me to believe they're going to plan for anything stronger, I guess.

11:58Well, I think he had a really good conversation with Justin about that idea of like, you know, the regulation coming in earlier. So I would refer people to politics, politics, politics from Friday, March 27th, because Andy and Justin got into that in a really good way. The one thing I'll push back on, though, is, again, I don't think this is going to turn into anything. That's fair. When your concern is, I don't think they'll push back on the provisions of this. I'm like, what provisions? This is a press release. That's my opinion. This is something that the administration can now say to the companies that it consulted with.

12:38See, we said the things you wanted us to say, knowing full well that there are so many other priorities for the Congress to get to that this won't turn into anything. And the other thing to remember is this administration is known for saying one thing and then letting something entirely different happen. You have in your notes here that Senator Blackburn introduced legislation to say that training your models on copyrighted works would not be fair use. A little blunt, if you ask me, but I wouldn't be shocked if that got momentum and passed, even though it's in direct contradiction to this framework, because that's the way things work right now.

13:21This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge, and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade-in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One. If you work in university maintenance, Grainger considers you an MVP. Because your playbook ensures your arena is always ready for tip-off.

13:59And Grainger is your trusted partner. offering the products you need all in one place from HVAC and plumbing supplies to lighting and more and all delivered with plenty of time left on the clock so your team always gets the win. Call 1-800-GRANGER, visit grainger.com or just stop by Grainger for the ones who get it done. Parles-tu français? Hablas espanol? Parli italiano? If you've used Babbel, you would. Babbel's conversation-based technique teaches you useful words and phrases to get you speaking quickly about the things you actually talk about in the real world. With lessons handcrafted by over 200 language experts and voiced by real native speakers, Babbel is like having a private tutor in your pocket.

14:43Start speaking with Babbel today. Get up to 55 % off your Babbel subscription right now at babbel.com slash acast. Spelled B-A-B-B-E-L dot com slash acast. Rules and restrictions May apply.

15:28Warby Parker store near you. Shipping, billing, admin, payroll, marketing. You're managing all the things. So why waste time sending important documents the old-fashioned way? Mail and ship when you want, how you want with Stamps.com. Print postage on demand 24-7 and schedule pickups from your office or home. Save up to 90 % with automated rate shopping. That's why over 1 million small businesses trust Stamps.com. Go to Stamps.com and use code PODCAST to try Stamps.com risk free for 60 days. I think you're absolutely right. This is effectively a press release right now. And I think you're totally fair to call that out.

16:09I think the thing that'll make me feel better about this is if we don't see a slowdown in the state AI laws. I think if the preemptions that they've put in this press release causes state legislations to deprioritize AI laws. Like a chilling effect against them. Yeah. Yeah. If it causes that, then we have a problem. If things proceed and they keep pushing forward with it, because that's really the only regulation and the only regulatory actions that we have at the moment. So it's the only test playground that we have to see what is the right level of pushback. Let me tell you, Ben. And I want to fully acknowledge that I think you're coming from a good place of like, I want there to be clear, good rules of the road that do our best to avoid harm from happening.

16:56And I think that I 100 % support you there. I think putting this out means that some states are going to accelerate their plans to regulate and and possibly and maybe not the best way because they rush to do it. But again, in the current in the old climate, if you ask me in 2012, I would say like, oh, yeah, this this could cause, you know, governors to to go, well, let's hold on. Let's see what Congress does with this framework. In today's climate, what this does is it tells a bunch of governors, push back more with your own state stuff and then use that to bash each other on social media and gain points with attack ads and hopefully get reelected.

17:40So I think this isn't even meant to dissuade states from passing their own legislation because they know that it will have the opposite effect, at least in some states. but from your mouth to some governor and state senators here uh california new york they're out there getting the word colorado's in that list as well sure i think illinois illinois uh oh yeah no those i i will leave it to the listener to go look up the governors of those states to understand why that might be but yes uh i i really i i barely i don't even think we included this in daily tech news show the day it was put out because i'm like great they said a lot of things call me when there's a bill call me when something has actually come out of this and then i will start paying attention to it that was definitely one thing i talked about with with justin a lot was this is far from the last time we're going to talk about it it's just the beginning of the conversation in many ways um i'm just skeptical uh of of of how it will play out and And I hope that you are the optimist here and that you are correct.

18:46I'm not even an optimist because because I think it's horrible that this is the situation we're in where a framework is meaningless. But I really do think that that is the situation we're in and that, yes, there will be actual bills created like the one Senator Blackburn created. They may or may not have any relationship to this framework in the future. And if that's optimism, I guess, then I have it. But I'm not even sure if it's pessimism, optimism or what. Maybe it's realism. Yeah, I've been accused of that before, too. Well, Andy, this was a great conversation. Thank you for taking a little time to hash over this.

19:24Really appreciate it. Before we go, what would you do? Let's make you the person who's like, Andy, we've left it to you. The government has designated you to create the situation. What kind of rules of the road do you think would be good to put in place? Oh, my God. This is the dream. I would love to be able to have some control over it. I would have had an AI oversight body with real authority. That would have been one of the mandates is go create budget and a group that can actually regulate this. and doesn't have another test they have to do. And go work with the governors of states that are creating legislation and use them as a test bed until you have federal regulations that can take them over and supersede them.

20:12And be more transparent around what are the requirements for an AI company using media in content training as a baseline for what licensing is going to look like in the future. Right now, all of that is being left up to the marketplace to decide. And I don't think there's enough regulation pushback in place. And I think you could find a way to influence it without overstepping. And I know it takes time to do that, but that's exactly what the state rules are there to sort of test out for the next few years. I am definitely with you on those last two. Transparency requirement and a deadline on the copyright question would be great.

20:52Even if the deadline isn't enforceable, just kind of putting it out there that we need an answer to this sooner than later. Right. It's a good way. I don't know if it needs explicit protection, but I'm with you on letting the states try things to see what works. I think in a calmer time when there's less performative actions, there's some real good faith efforts that could be made to say like, well, let's try it and see what effect it has see if it actually causes good um i don't know if i want another oversight body uh to do this or but i also am having a hard time coming up with an existing body that is competent to do this so i'm i'm a little fuzzy i'm not against you on that one it's more like yeah i'm not i'm not sure i would have to do some more thinking on that uh for it but i i i'm i'm for electing andy uh as ai oversight czar well i don't know that it fits in cleanly anywhere else like i tried to imagine FCC?

21:50Is it SEC? Is it where does it fit in? And I don't know that there is one. If we can find a home for it, I'm from I'm for that. But again, they probably need budget and a mandate to go take it over. Yeah, almost like what's who's in charge of product safety? The FDA doesn't seem like the right one for this. Yeah, no. But it's yeah, something. But you're right. I mean, I don't know where this cleanly fits into our current places. And if we can find a home for it, great. Again, I would rather think about technology than politics. National Highway Transportation and Safety Administration, since this stuff's going in our cars, maybe they could take the lead on this.

22:30Ooh. I don't know. Crazier things. I like it. Feedback at DailyTechNewsShow.com. If you have the solution, we'll submit it to our AI czar, Andy Beach. Thank you for joining us. If folks want to see more of your thoughts on this, because you really do think about this stuff and you do great work on engines of change, where can they go? Yeah. So engines of change.ai is my sub stack. I write a lot about this over there right now. I'm actually in the middle of writing a new series that I refer to as the Inheritance Series. And it's about how the infrastructure that creates media is starting to contextually understand it.

23:09And what are the implications of that? Fantastic. Thank you, Andy. And thanks, everyone, for joining us. We'll be back with regular news episodes on Monday. Talk to you then, everybody.

23:32The DTNS family of podcasts. Helping each other understand. Diamond Club hopes you have enjoyed this program.

23:44This episode is brought to you by Capital One. Capital One's tech team isn't just talking about multi-agentic AI. They already deployed one. It's called Chat Concierge and it's simplifying car shopping. Using self-reflection and layered reasoning with live API checks, it doesn't just help buyers find a car they love. It helps schedule a test drive, get pre-approved for financing, and estimate trade-in value. Advanced, intuitive, and deployed. That's how they stack. That's technology at Capital One.

24:41To get it done.