The Little Tech Agenda for AI

0:00:05 There have been these big institutional players in DC in the state capitals for a very long time.
0:00:09 There wasn’t anyone who was actually advocating on behalf of the startups and entrepreneurs,
0:00:14 the smaller builders in space. They’re trying to build models that might compete with Microsoft
0:00:19 or OpenAI or Meta or Google. For those companies, what are the regulatory frameworks that would
0:00:23 actually work for them as opposed to making that competition even more difficult than it already is?
0:00:28 Regulate use, do not regulate development, somehow is interpreted as do not regulate.
0:00:36 I actually can’t think of a single example across the portfolio in which we are arguing for zero regulation.
0:00:42 Who’s speaking up for startups in Washington, DC? Today, I’m joined by Matt Peralt,
0:00:48 Head of AI Policy at A16Z, and Colin McKeown, Head of Government Affairs at A16Z,
0:00:52 to talk about the Little Tech Agenda, a framework designed to ensure that regulation doesn’t just
0:00:57 work for the giants, but also for the five-person teams trying to build the next breakthrough. Their
0:01:04 approach, regulate harmful use, not development. We’ll cover federal versus state rules, open source,
0:01:08 export controls, and what smart preemption could look like. Let’s get into it.
0:01:13 Colin, Matt, welcome to the podcast.
0:01:15 Thanks so much. Thanks for having us.
0:01:19 So there’s a lot we want to get into around AI policy. But first, I want us to take a step back
0:01:25 and reflect a little bit. We had publicly announced the Little Tech Agenda in July of last year.
0:01:30 There’s a lot that’s happened since. When we first take a step back, Colin, and talk about what is the
0:01:32 Little Tech Agenda and how did it come to be at the firm?
0:01:36 Yeah. I mean, look, a ton of credit to Mark and Ben for having sort of the vision on this.
0:01:43 I think certainly when I first started here, I arrived, we started advocating on behalf of
0:01:50 technology interest, technology policy. And I think what we realized was there have been these big
0:01:54 institutional players that have been in DC in the state capitals for a very long time.
0:01:59 Some of them have done a lot of really good work on behalf of the entire tech community.
0:02:05 But there wasn’t anyone specific who was actually advocating on behalf of what I think what we call
0:02:10 little tech, which I think in my mind are the startups and entrepreneurs, the smaller builders
0:02:15 in the space. And I think beyond that, what we realized was, well, they’re not always 100% aligned
0:02:20 with what’s going on with the big tech folks. And that’s not necessarily always a bad thing or a good
0:02:25 thing. But I think that was the whole impetus of this. You know, how are we going to think about
0:02:29 positioning ourselves in DC and the state capitals in terms of our advocacy on these issues?
0:02:36 And how do we differentiate in between sort of the big tech folks who come with their certain
0:02:41 degrees of baggage? And the smallest is small. From the left and the right, right? And the smallest is
0:02:44 small. So that was really sort of the basic impetus of this. For me, it was actually sort of
0:02:49 almost a recruiting vehicle. So when it hit in July, I was not yet at the firm. I started in November.
0:02:54 And when I first read the agenda, it sort of transformed the way that I looked
0:02:58 at the rooms that I would sit in, where there would be policy conversations, where all of a sudden,
0:03:02 you could see essentially an empty seat. And Little Tech’s not there. You know,
0:03:06 there would be conversations where people would say, and in this proposal, we want to add this
0:03:09 disclosure requirement. And then we’ll have companies do a little bit more and a little bit more.
0:03:13 And when you’ve read the Little Tech agenda, all of a sudden, you start thinking, how is this
0:03:17 going to work for all the people who aren’t in the room? And so for me, the question like thinking
0:03:22 about coming into this role in the firm was, is this a voice? Is this a part of the community I want to
0:03:26 advocate for and think about? And when you start looking at the policy debate from a perspective
0:03:30 of Little Tech, and you see how many of the conversations don’t include a Little Tech
0:03:34 perspective, it comes, from my point of view, is very compelling to think about how can I advocate
0:03:36 for this part of the internet ecosystem?
0:03:36 Right.
0:03:41 And Kyle, why don’t you outline some of the pillars of the Little Tech agenda, or some of the things that
0:03:45 we focus the most on, and maybe how it differentiates from sort of big tech more broadly?
0:03:50 Yeah. I mean, well, I mean, just from a firm perspective, right? Obviously, we’re verticalized.
0:03:53 You know, we all live and breathe this. And I think that that’s been very, very competitive
0:03:57 for us on the business side. But I also think it’s very competitive of us on the policy side,
0:04:03 too, right? Obviously, MAD leads our AI vertical and that’s sort of our AI policy lead. We have a huge
0:04:07 crypto effort. We have a major effort around American dynamism. And then this is sort of defense
0:04:11 procurement reform, which is something that the United States is needed forever and ever.
0:04:16 We have, you know, other colleagues who work on the bio and health team. And they’re fighting on
0:04:21 behalf of, you know, FDA reform, everything from PBMs. There’s a whole vertical there that they’re
0:04:27 working on. We’re working a lot on fintech-related issues. And then, you know, just like classic
0:04:34 tech-related sort of internet entrepreneurs coming up. What does that relate to? There’s a lot of tax
0:04:38 issues that come along with it. And then, of course, obviously, there are the venture-specific
0:04:42 things that we have to deal with. But look, I think I try and think about this from a basic
0:04:48 point of view, which is just like, if you’re a small builder, what are the things that should
0:04:52 differentiate you between someone who’s a trillion-dollar company and you have hundreds of thousands of
0:04:58 employees, right? If you’re five people and you’re in a garage, how are you supposed to be able to
0:05:03 comply with the same things that are built for a thousand-person compliance teams? Like, it’s just not
0:05:07 the same thing. Right. And like, there are categories and categories that, you know, Matt and I are
0:05:12 dealing with on a regular basis. But that’s probably the main pillar, which is five-person
0:05:17 versus trillion-dollar company, not the same thing. It’s made my job actually really hard in certain
0:05:23 ways since I started at the firm because the kinds of partners that you want within our portfolio
0:05:28 often don’t exist in that, like, a lot of the companies don’t have a general counsel.
0:05:32 Yeah. They don’t have a head of policy. They don’t have a head of communications. And so the
0:05:37 kinds of people who typically sit at companies thinking all day about, like, what is this state
0:05:42 doing in AI policy? What is this federal agency doing in terms of rulemaking? They’re not at startups that
0:05:48 are just a couple of people and engineers trying really hard to build products. Those companies face
0:05:53 this incredibly daunting challenge. I mean, it seems so daunting for someone like me, like non-technical,
0:05:57 and I’ve never worked at a startup. If they’re trying to build models that might compete with
0:06:02 Microsoft or OpenAI or Meta or Google, and that is unbelievably challenging in AI. You have to have
0:06:07 data. You have to have compute. There’s been a lot written about the cost of AI talent recently. It’s
0:06:12 incredibly, incredibly daunting. And so the question that Colin and I talk about all the time is for
0:06:17 those companies, what are the regulatory frameworks that would actually work for them as opposed to making
0:06:21 that competition even more difficult than it already is? Yeah. Well, yeah, one of the principles I’ve heard
0:06:26 you guys hammer home is we want a market that’s competitive where startups can compete. We don’t
0:06:30 want a monopoly. We don’t want even oligopolies, you know, a cartel-like system. And that doesn’t
0:06:36 mean no regulation because that can, as we’ve seen, that could be destabilizing too. But it means smart
0:06:41 regulation that enables that competition in the first place. Yeah. So I think one of the things that’s been
0:06:47 surprising to me to learn about venture is the time horizon that we operate in. So our funds are 10-year
0:06:54 cycles. So we’re not looking to spike an AI market tomorrow and have a good year or a good six months or a good
0:07:00 two years. We’re looking to create vibrant, healthy ecosystems that result in long-run benefits for people and long-run
0:07:07 financial benefits for our investors and for us. And that means having a regulatory environment that facilitates
0:07:14 healthy, good, safe products. It doesn’t mean, like, if people have scammy, problematic experiences with AI
0:07:18 products, if they think AI is bad for democracy, if they think it’s corroding their communities,
0:07:24 that’s not in our financial incentive. That’s not good for us. And so that really animates the kind of
0:07:31 core component of the agenda, which is not trying to strip all regulation, but instead focusing on
0:07:35 regulation that will actually protect people. And we think that there are ways to do that without making it
0:07:39 It’s making it harder for startups to compete. Yeah, to Matt’s good point, I walk into a lot of lawmaker offices.
0:07:44 You know, it sounds like I’m pitching my book, but I genuinely say, like,
0:07:50 Our interests are aligned with the United States of America’s interests. Yeah. Because the people that we’re funding
0:07:55 are on the cutting edge. They’re the people who are going to build the companies that are going to drive the jobs.
0:08:00 They are going to drive the national security components that we need. And they’re also going to drive the economy.
0:08:07 And like, we want to see them build over a long time horizon. And like, that is exactly how we should be
0:08:10 building policy in the United States. Of course, like half the offices I walk into, like, all right,
0:08:17 great. Get that guy out of here. 99.9% of people we talk to think that all we want is no regulation.
0:08:23 And yeah, despite like writing extensively, both of us writing, speaking extensively about the importance
0:08:27 of good governance for creating the kind of markets that we want to create. If Colin can speak more to
0:08:31 it in crypto, I’ve learned a lot from our crypto practice, because the idea there is you really need
0:08:36 to separate good actors from bad actors and ensure that you take account for the differences. And it’s
0:08:42 true in AI as well. If we don’t have safe AI tools, if there is absolutely no governance,
0:08:46 that’s not going to create a long run healthy ecosystem that’s going to be good for us and
0:08:50 good for people throughout the country. I actually can’t think of a single example
0:08:55 across the portfolio in which we are arguing for zero regulation.
0:09:00 The core component of our AI policy framework, which was developed before my time, I wish I
0:09:05 could take credit and I can’t, is focused on regulating harmful use, not on regulating development.
0:09:11 And that sentence, regulate use, do not regulate development, somehow is interpreted as do not
0:09:16 regulate. And people just omit for some reason, the part that we focus on, on focusing on regulating
0:09:21 harmful use. And that in our view is robust and expansive and leaves lots of room for policymakers
0:09:25 to take steps that we think are actually really effective in protecting people. So regulating
0:09:29 use means regulating when people violate consumer protection law, when they use AI to violate
0:09:34 consumer protection law, or when they use AI in a way that violates civil rights law at the state and
0:09:39 federal level or violating state or federal criminal law. So there’s an enormous amount of action there
0:09:45 for lawmakers to seize on. And we really want that to be like an active component of the governance agenda
0:09:49 that we’re proposing. And for some reason it’s all passed over and the focus is just on don’t
0:09:53 regulate development. I don’t exactly understand why that ends up being the case.
0:09:57 – Easy headline. – So there’s been a lot that’s happened in AI policy and I want to get to it,
0:10:02 but first, perhaps Matt, you can trace the evolution a bit over the last few years. I believe there was
0:10:05 a time where we were like pattern matching with social media regulation a bit. Why don’t you trace
0:10:08 some of the biggest inflection points, kind of the debates over the last few years, and we’ll get to
0:10:13 today, maybe, Carl? – I think we have to play a little bit of history. And I want to get to,
0:10:17 you know, sort of a point that I think is the really critical point of what we’re all facing here.
0:10:23 for us, for me, I would say from a policy and government affairs perspective, this conversation
0:10:31 started early 2023. That was sort of like the kickoff of the gun. It sort of puttered along and
0:10:39 became more and more real over time. But in the fall of 2023, so almost exactly to the day two years ago,
0:10:44 there was a series of Senate hearings in which, you know, some major CEOs from the AI space came and
0:10:51 they testified. And I think that the message that folks heard was, one, we need and want to be
0:10:56 regulated, which I think maintains that truth today. That’s obviously, you know, what Matt and I are
0:11:05 working on on a regular basis. But I think included in some of that testimony was a lot of speculation
0:11:12 about the industry that led to and sort of absolutely jump-started this whole huge wave
0:11:19 of conversation around the rise of Terminator, you know, go hug your families because we’re going to
0:11:25 all be dead in five years. And that spooked Capitol Hill. I mean, they absolutely freaked out about it.
0:11:29 And look, rightfully so. You have these really important, powerful people who are building this
0:11:32 really important, powerful thing. And they’re coming in, they’re going to tell you that, you know,
0:11:37 everyone’s going to die in five years, right? That’s a scary thing for people to hear. And,
0:11:43 oh, by the way, we want to be regulated. Which, you know, look, that starting gun,
0:11:50 I think, moved us in hyperspeed into this conversation around, how do we lock this down? How do we regulate
0:11:55 it very, very, very quickly? I think that led to the Biden executive order, which, you know,
0:12:03 we’ve publicly sort of, you know, denounced in certain categories. Um, that executive order led
0:12:08 to a lot of the conversation that I think we’re having in the states, a lot of the, you know,
0:12:15 sort of bad bills that we’ve seen come through the states. Um, and I think it also, um, led to a number
0:12:20 of federal proposals that we’ve seen that have not been very well thought through also. And look,
0:12:24 you know, I think people are kind of sitting around, they’re like, oh, well, you know,
0:12:29 was it just like, you know, some testimony from the CEOs that did this? And the answer to that is no.
0:12:35 You know, from my point of view, um, and look, you know, they deserve a lot of credit. I think the
0:12:43 effect of Altruist community for 10 years backed by large sums of money, um, were very, very effective
0:12:52 at influencing think tanks and nonprofit organizations in DC and the state capitals to sort of push us in
0:12:59 a direction where, um, people are very fearful about the technology. And that has, that has shaped,
0:13:05 significantly shaped the conversation that we’re having throughout DC and the state capitals and
0:13:10 candidly on a global stage. You know, the EU acting, the EU AI act, you know, we’re public on that.
0:13:16 There’s a lot of very, very problematic provisions in there. All of this banner of safetyism came from
0:13:21 this 10 year headstart that these guys have had. So when I always, you know, that that’s kind of a bit
0:13:28 of the history, but sort of as an aside to this, I always, I always just have to smirk or, you know,
0:13:33 smile to try and laugh it off. But I mean, when people are writing these articles about the fact that
0:13:38 the AI industry is, you know, pumping all this money into the system, certainly like I’m not suggesting
0:13:42 that there’s not money in the system. We’re obviously active on the political and policy side.
0:13:47 We’re, you know, we’re not hiding that, but it is dwarfed by the amount of money that is being
0:13:52 spent and has been spent over a 10 year window. And, and candidly, I mean, the reason that Matt
0:13:57 and I have jobs is because we are playing catch up. We are here to try and make sure that people
0:14:02 understand what is actually going on in this conversation and be a counterforce to this,
0:14:08 this group of people and, and this idea, this ideology that has been here for a long period of time.
0:14:13 So that’s, you know, that, that’s kind of the briefer on this. Yeah. I mean, and, and companies,
0:14:19 I think we’re ready to consider some policy frameworks that, that I think we’re probably
0:14:24 really going to be challenging for the AI sector in the long run. Right. And I think that’s because
0:14:33 I w I was at, um, Meta then Facebook starting in 2011 and through 2019. And so after really like 2016,
0:14:37 there was aggressive criticism of tech companies and the general framing is like, um, you’re not
0:14:43 being responsible and regulation needs to catch up. You, um, governance of social media is behind where
0:14:49 the products are. Um, and whatever you think about that, that was really the kind of strong view in the
0:14:54 ecosystem that like governance has allowed the lack of governance has allowed problematic things to happen.
0:15:02 And so I think when AI was starting to accelerate and, um, and, and you had certain sort of prevailing
0:15:06 political interests, I think that were driving the conversation, companies rushed to the table.
0:15:11 And I think it was a group of five, three, five, seven companies who went into the White House
0:15:16 and negotiated voluntary commitments. Yeah. I mean, we don’t even have to make the argument about the
0:15:21 importance of representing little tech in, when you see that there is a set of companies who
0:15:28 negotiated an arrangement for what it would look like to build AI at the frontier with all current
0:15:32 developers who weren’t those companies and all future startups not represented at the table.
0:15:39 Yeah. I think that is why, like, we started to think about the value of having more dedicated
0:15:43 support around AI policy because clearly the views of little tech companies aren’t represented in the
0:15:50 conversation. Yeah. Well, I mean, let me, let me just add one thing to this. And I, I, I, it’s Mark and Ben’s
0:15:55 story. They’ve told it many times. I was in the meeting as well, you know, and, and like, you know,
0:16:00 everything they’ve said has been a hundred percent true and accurate, but there was a, there was a
0:16:05 prevailing view by very, very powerful people of the previous administration
0:16:13 that this was going to be only two or three major companies able to compete in the AI landscape.
0:16:20 And because that was the case, they needed to be basically locked down and, and put in this incredibly
0:16:25 restrictive view from a policy and regulatory perspective. And that was going to be kind of like
0:16:30 this, this entity that was kind of like our arm of the government. And, and I think that that was
0:16:37 the most alarming thing that I think we had heard from the administration on top of an incredibly
0:16:42 alarming series of events that happened on the crypto side, um, including sort of wanting to eradicate it
0:16:49 off the face of planet. It seemed like, so I, I think that that all led to kind of the position that we’re in
0:16:54 now and certainly like Matt’s hiring and the thing, you know, like us building out the team, et cetera.
0:17:01 So that narrative is clearly like a very alarming, maybe the most alarming version of this. But even
0:17:04 since I’ve been in this role, I’ve heard other versions of it where people will say, oh, don’t
0:17:08 worry about this framework. It just applies to three or five companies, or it just applies to five to seven
0:17:13 companies. And I think they mean that to provide comfort to us. Like, oh, this isn’t going to cover a lot of
0:17:18 startups, but the view of the AI market where there are only a small number of companies building at the
0:17:22 frontier is not the, that’s not the vision for the market that we have. We want it to be competitive
0:17:27 and diverse at the frontier and the policy ideas that were coming out of the period that Colin’s
0:17:32 talking about were dramatically different from where they are today in a way that I think like some people
0:17:38 have even like lost sight of exactly where we were a couple of years ago. There were ideas being proposed
0:17:44 by not just government, but industry to require a license to build frontier AI tools and for it to
0:17:48 be regulated like nuclear energy, not just a historic for software development.
0:17:53 Yeah, right. Unprecedented. Yeah. Yeah. And for it to be regulated like nuclear energy with like an
0:18:01 international style nuclear, like, sorry, an international level nuclear style regulatory
0:18:05 regime to govern it. And we’ve moved, like, no matter what you think about the right level of
0:18:09 governance, there are not a lot of people now who are saying what we need as a licensing regime where
0:18:13 you literally apply for permission from the government to build the tool. But that wasn’t
0:18:17 that far in the rearview mirror. Yeah. And look, and we’re also talking about bans on open source.
0:18:22 Yeah. I’m like, we’re still kicking around that idea at the state level. And, and look, I, I, it all,
0:18:27 you know, look for us who live and breathe the tech stuff on a daily basis, this, this is, you know,
0:18:32 this sounds insane and crazy. But let me, you know, like just to make it a little bit more real,
0:18:42 like the nuclear policy in the United States has yielded two, three new nuclear power plants in a
0:18:47 50 year period since these organizations have been started. And look, like you can, some people are
0:18:51 pro nuclear or some people are anti nuclear. I, I don’t want to get into that debate. The point though
0:18:57 is, is that that was not the intended policy of the United States of America. That was the effect of putting
0:19:04 together this agency and what has come from that. And I think, you know, look, if we do the same
0:19:09 thing to AI, had we done the same thing in AI in that period of time, then you don’t have the medical
0:19:14 advancements. You don’t have the breakthroughs. You don’t have all of the things that come from this
0:19:19 that are incredible. But beyond that, we lose to China. Yeah.
0:19:24 Full stop. You lose to China. And then our greatest national security threat becomes the one who has
0:19:28 the most powerful technology in the world. Right. And I think, I think the early concern
0:19:31 on the open source was that we would be somehow giving it to China, but then we’ve seen with deep
0:19:36 seek, et cetera, that they just have it anyways. Yeah. Yeah, exactly. Right. Exactly. You know,
0:19:40 the idea that we could lock this down, I think, I think, you know, I mean, Mark and Ben have talked
0:19:45 about this. I mean, I think they’ve debunked that a number of times. Yeah. Just understand was for the
0:19:50 previous administration, what was their calculus? Was it that they were true believers in the fears?
0:19:54 Was it that there was some sort of political benefit to having the views that they had,
0:19:58 especially on the crypto side? I don’t understand what’s what is the constituency for anti crypto
0:20:03 stance? How do you make sense of of sort of the players or the intentions or motivation? Just
0:20:08 understand sort of the calculus there. Yeah. You know, I mean, look, I think that that’s a really,
0:20:12 I think that’s a really hard one to answer and I’m not sure I can pretend to be completely in their
0:20:18 minds. I think there’s a couple of different competing forces here. Like one is, you know,
0:20:24 what are the constituencies that support the sort of that administration? What are the constituencies
0:20:31 that support that side of the aisle? And I think that especially over the last 10 to 15 years,
0:20:35 it has been very, very heavy, heavy focus on consumer fit, consumer safety, which I think,
0:20:39 look, a very important thing. And we’re obviously in alignment on that. I think everyone should be
0:20:42 in alignment, have to protect consumers, have to be able to protect the American public.
0:20:50 But I think that, um, a lot of that conversation has been weaponized. I think that it is, it is a
0:20:56 big time moneymaker. I think a lot of these groups either get backing from very, very wealthy,
0:21:03 special interest or they are small dollar fundraising off of quick hits like, you know,
0:21:08 AI is coming for your jobs, donate $5 and we’re going to, you know, and we’ll make sure that we
0:21:13 take care of this in Washington for you. And, you know, like pretty easy, you know,
0:21:19 it’s a pretty easy manipulation tactic, you know, it’s used like from a bunch of people, but I think
0:21:25 that that’s like a very, that held very seriously true. Right. And I think, um, you know, the other
0:21:31 thing here is that I think, um, personnel is policy. It’s the old saying, personnel is policy. And I think
0:21:37 a lot of the individuals that, um, were in very senior decision-making roles within that White House and
0:21:43 that administration came from this sort of consumer protection background where they’ve seen this,
0:21:49 that was their constituency. They were put in this position to come after private enterprise. Like,
0:21:54 you know, that was, that was the, that was the goal. Like, there’s this whole idea out there,
0:21:59 I think among some of those folks that, you know, Senator Warren has, has, you know, proposed this many
0:22:06 times is, is like, if you’re not getting, you know, if you’re not going after and getting people on a
0:22:11 regular basis in the private sector, then you’re not working hard enough. And I, and like, I, I just,
0:22:17 you know, I think that that, that is, is probably like the second thing. And then like the third is
0:22:26 just, we’re at this very weird moment where being a builder and being in private enterprise is, is a bad
0:22:32 thing to some policymakers. It’s not, you know, you’re not doing good because you’re earning a profit.
0:22:37 And, you know, they certainly won’t say that, but the activities and the things that they’re doing
0:22:44 are a hundred percent alive with that, that type of idea. So I, you know, I, I think that’s the basic
0:22:45 crux of it.
0:22:54 I, I think the, the, the things that motivated that approach were done in good faith. And I think, I think
0:22:59 it’s what you alluded to earlier, which was like, I don’t share this view, but there are a lot of people
0:23:05 who believe that social media is poorly regulated. And that because policymakers were asleep at the wheel,
0:23:11 we woke up at some point, I don’t know, sometime in the 2014 to 2018 period, and realized that we had technology
0:23:16 that we thought was actually not good for our society. And I think that whether or not you think
0:23:21 that that’s true or not, that I think that was, that has been a widely held view. It’s a, it’s a held
0:23:25 view on the right and on the left, it’s a bipartisan view. And so I think when this new technology came on
0:23:30 the scene, this was a do over opportunity for policymakers, right? Like we can get this right,
0:23:35 when we didn’t get the last thing right. And so I understand that motivation. It makes a lot of
0:23:42 sense. I think the thing that we, that we strongly feel is the set of policy ideas that came out of that
0:23:48 like good faith belief, were not the right policy ideas to either protect consumers or lead to a
0:23:54 competitive AI market. Like some of many of the politicians who are pushing concepts that would have
0:24:01 really put a stranglehold, I think on AI startups and would have led to more monopolization of a
0:24:05 market that already tends toward monopoly because of the high barriers to entry already. Um, those
0:24:09 politicians three years before had been talking about how problematic it was that there wasn’t more
0:24:13 competition in social media. Um, and then all of a sudden they’re behind, you know, a licensing regime,
0:24:17 which is not, I don’t think there’s much economic evidence that licensing is pro-competitive.
0:24:23 It typically is the opposite. The disagreement is less with the core feeling. Like we want to protect
0:24:28 people from harmful uses of this technology and more from the policy concepts that came out of that
0:24:33 feeling that we think would have been disruptive in a problematic way to the future of the AI market.
0:24:39 Yeah. Anecdotally, it seemed from, from afar that some of the concerns early on were almost,
0:24:44 you know, to match social media, like around disinformation or even like DEI concerns. Um,
0:24:48 and then, you know, people were trying to sort of, uh, make sure the models were incompatible with,
0:24:52 what about with sort of the sort of, you know, um, speech regime at the time, but then it’s kind of
0:24:56 shifted to, Oh wait, no, is, is, is there more existential concerns around jobs or, or is AI even
0:25:02 like nukes in, in the sense of like, and even people doing harm or AI itself doing harm, but it seemed to,
0:25:05 to escalate a bit and, you know, uh, maybe aligned with that testimonial that you alluded to.
0:25:10 I, I, I experienced it as feeling like the gold posts always move. And one of the things that I said,
0:25:15 like that, I said that I started asking people when I was really trying to settle into this
0:25:20 regulate use, not development policy position is what do we miss? Like if we regulate use primarily
0:25:24 using existing law, what are the things that we miss? And I haven’t gotten very many clear answers
0:25:31 to that, right? Like you can’t do illegal things in the universe and you also can’t use AI to do illegal
0:25:35 things. And typically when people list out the set of things that they’re most concerned about
0:25:40 with AI, it are, they’re typically things that are covered by existing law. Not, probably not
0:25:46 exclusively. Right. But primarily. And so that at least seems like a, a good starting point. Some of
0:25:50 the other issues that I think are like understandably ones that we should be concerned about have a range
0:25:55 of different considerations associated with them. Like the, like if you’re concerned about misinformation
0:25:59 or like speech that you think might not be true or might be problematic, there are significant
0:26:03 constraints on the government’s ability to regulate that. Yeah. The first amendment imposes
0:26:08 pretty stringent, uh, restrictions. And I think for very good reason, because you don’t want the
0:26:12 government to dictate the speech preferences, policies of private speech platforms for the most
0:26:18 part. And so, um, so that those issues might be concerns, but they’re not necessarily areas I think
0:26:23 where you want the government to step in and take strong action. Um, and so there, I think there are
0:26:27 things that we should probably do as a society to try to address those issues, but regulate government
0:26:31 regulation maybe isn’t the primary one. And again, and most of the things that people are most
0:26:36 concerned about, like real, real use of the technology for clear cognizable real world harm,
0:26:43 existing law typically covers it. I have a theory on this. So I think everything that Matt just said is,
0:26:46 is spot on, but, but, you know, like then, then you’re kind of sitting around and you’re kind of
0:26:53 scratching your head. It’s like, okay, well, if use covers it and there hasn’t been, you know, a, a very
0:27:00 incredibly fair rebuttal onto why use is not enough in, in terms of focus on, on the policy and regulatory
0:27:05 side, what’s, what’s the answer. I think we’re, we’re experiencing sort of this, I don’t know if
0:27:10 it’s phenomena, but we’re experiencing this pattern on the crypto side too, which is, which is we’re
0:27:16 having a very, very spirited debate on the crypto side of things on how to regulate sort of these
0:27:19 tokens and how do you launch a token in the United States as a security or as a commodity. And this
0:27:23 is sort of this age old debate that’s, you know, plagued securities, traditional securities laws for
0:27:30 years, but also certainly with crypto industry. But what we have found is there are, there are a
0:27:37 number of people who have entered this debate who are actually trying to get at the underlying
0:27:42 securities laws. Like they, they want to reform securities laws. They don’t want to reform crypto
0:27:48 laws that involve securities. And this is their only venue by which they can enter that conversation.
0:27:54 Because we’re not having, there, there’s no will from the Congress or from policymakers to go and
0:28:00 overhaul the securities laws right now. You know, it’s just not there. But what is moving is crypto.
0:28:04 So people, you know, there are all these people that are now trying to enter this debate and like,
0:28:08 oh, we should relook at this and like, well, this doesn’t have anything to do with it. We shouldn’t
0:28:12 be entering this conversation yet. They’re still pushing. Right. And that’s kind of muddy the water.
0:28:16 I think a very similar thing is actually happening on the AI side, which is, you know,
0:28:22 there are a number of members of Congress that feel like, well, we missed it on the 96 Telecom Act.
0:28:29 Like that wasn’t, we didn’t do good enough around then. So we need to rewrite the wrongs
0:28:35 through the venue of an AI policy conversation. Right. Because if you, if you think about it,
0:28:41 right. Assuming that use doesn’t go far enough for someone. Right. And this is the same conversation
0:28:45 that we’re having in California right now or in Colorado right now, if uses does not go far enough.
0:28:51 Okay. Well, then it would be really, really simple if you could have a privacy conversation around this,
0:28:55 if you could have an online content moderation conversation, an algorithmic bias conversation
0:29:00 around it, you could do all of that, wedge it through AI. And then assuming AI is actually going
0:29:06 to be the thing that we all think it’s going to be. Now you’ve put basically a regular regulatory
0:29:10 funnel on the other side, like you’ve put a mesh screen where everything has to run through AI,
0:29:13 and therefore it runs through this regulatory proposal you put together.
0:29:17 Yeah. The, the, the thing that I’ve really been wrestling with in the last few weeks
0:29:23 is whether those kinds of regimes are actually helpful in addressing the harm that they purport
0:29:27 to want to address. And Colorado is a really good example. So there are all these bills that have
0:29:31 been introduced at the state level. Colorado is the only one that’s passed so far. It’s set up this,
0:29:37 this regime where you basically have to decide, are you doing a high risk use of AI or a low risk use
0:29:41 of AI? And this would be for startups that don’t have a general counsel, don’t have a head of policy,
0:29:45 can hire an outside law firm to figure it out high risk, low risk. And then if you’re high risk,
0:29:49 you have to do a bunch of stuff, usually impact assessments, sometimes audit your technology
0:29:54 to try to anticipate, is there going to be bias in your model in some form, which maybe an impact
0:29:58 assessment helps you figure that out a little bit, but it’s probably not going to eliminate
0:30:03 bias entirely. It certainly isn’t going to like end racism in our society.
0:30:12 Colorado is now their governor, their attorney general have put pressure on the legislature to
0:30:16 roll back this law because they think it’s going to be problematic for AI in Colorado. And so there
0:30:20 was just a special session there to consider various different alternatives. One of the alternatives that
0:30:28 was introduced proposed codifying that the use of AI to violate Colorado’s anti-discrimination statute
0:30:33 is illegal. That’s consistent with the regulate harmful use framing that we’ve talked about. And it’s
0:30:39 instead of having this like amorphous process where maybe you address bias in some form, maybe you don’t,
0:30:45 this goes straight at it. It’s not a bank shot. It goes straight at it, where if someone uses AI in a way that violates
0:30:50 anti-discrimination law, that that would be, that could be prosecuted. The attorney general could
0:30:56 enforce. And I don’t, I still don’t understand why that approach is not, is somehow less compelling than
0:31:00 this complex administrative paperwork approach. I think it’s kind of the reason that Colin’s describing,
0:31:06 which is like people want another, a different bite at the apple of bias, I suppose. But it’s not
0:31:11 clear to me that it’s actually the best way to effectuate the outcomes that you want, as opposed to
0:31:15 just criminalizing or creating civil penalties for the harm that you can see clearly.
0:31:23 It’s also, I mean, in policymaking and bill writing, it, it, it’s really, really easy to come up with bad ideas.
0:31:23 Yeah.
0:31:28 It’s easy, right? Because they’re not well thought through. The first thing comes to your head, someone publishes a paper on
0:31:35 something, here we go. It takes real hard work to get something that actually works. And then it’s even harder to
0:31:39 actually go through a political and policy negotiation with a diverse set of stakeholders and actually land the plane on
0:31:44 something. Yeah. I think that’s part of the reason that people think that we are anti-governance,
0:31:49 because when we, I mean, Colin, again, he lived this history, I’m coming in late to it. But like,
0:31:55 as we were ramping up our policy apparatus, these were the ideas in the ecosystem, licensing, nuclear
0:32:02 style regulation, like flops, threshold based disclosures, really complicated transparency regimes,
0:32:07 impact assessments, audits, which are a bunch of ideas that we think are not going to help protect
0:32:13 people and are going to make it really hard for low resource startups. And so we’ve been trying to say,
0:32:17 no, no, no, don’t do that. And so that sounds like deregulate. But, but for whatever reason,
0:32:22 it’s been hard so far to shift toward like, here’s another set of ideas that we think would be compelling
0:32:25 and actually protecting people and creating stronger AI markets.
0:32:32 Right now we don’t see, you know, terrorists or criminals being aided, you know, 1000X with AI
0:32:36 and in performing terrorism or crime. Like when I ask people, like, what are you truly scared about?
0:32:40 Like, give me a concrete scenario. People, you know, they’ll be like, oh, what about like bioterrorism or
0:32:45 something? Or what about, you know, cybersecurity, you know, theft or something? We seem very far away from
0:32:51 that. Is there any amount of development at, you know, in the next few years, any amount of breakthroughs
0:32:58 where you, where you might say, oh, you know, maybe use isn’t enough or, or do we think that that will always be a?
0:33:04 I think it’s conceivable. I mean, and I think we’ve been open about that. Like we, we, we think existing law is a good
0:33:10 place to start. It’s probably not where we end. So Martin Casado, one of our general partners, wrote a great piece on
0:33:15 marginal risk in AI, basically saying like when there’s incremental additional risk that we should
0:33:19 look for policy to address that risk. And so the situation you’re describing, I think might be that.
0:33:24 I think what you’re getting at is a really important question about just potential significant harms that
0:33:31 we don’t yet contemplate. We get asked often about our regulate use, not regulate development framework.
0:33:35 Are you just saying that we should address issues after they occur? And I understand why that’s a concern.
0:33:41 Like there might be future harms. Um, and wouldn’t it be nice if we could prevent them in advance,
0:33:47 but that is how our legal system is designed. And typically when you talk to people about ways that
0:33:54 you could try to address potential criminal activity or other legal violations ex ante before they occur,
0:33:59 that’s really scary to people. Like Eric, what if we just learned a lot of information about you and then
0:34:03 predicted the likelihood that you might do something unlawful in the future. And if we think it’s exceeded a
0:34:07 a certain threshold, then we’re going to go and try and take action against you before you’ve done it
0:34:12 so that we can prevent future crime that you’re laughing because it’s laughable. We, we don’t want
0:34:19 a kind of ex ante surveillance, um, both because it feels invasive, but also because it, it’s, it,
0:34:24 it often is ineffective. Like you might, it might, we might run some tests that shows that maybe you’re
0:34:29 likely to pre be predisposed to some kind of criminal activity, but we don’t know until you’ve done it,
0:34:34 that you’re going to do that. You’ve done it. And so, um, I think that kind of approach, again,
0:34:41 I think it’s motivated by a really valid concern and a valid desire to prevent harm.
0:34:45 What if we could prevent harm before it’s occurred? The challenge is the regulatory framework, I think
0:34:50 probably won’t do that. It probably won’t have the effect of preventing harm. And there are all these costs
0:34:53 associated with it, mainly from our perspective, inhibiting startup activity.
0:35:00 Yeah. Mark, um, once told me on a podcast, he told me his joke, which is, uh, man goes to the
0:35:05 government, uh, you know, I go to the government because, uh, I have this big problem. Uh, now I
0:35:12 get a lot of regulation. Now I have two problems. Okay. Let’s talk about the state of AI policy today.
0:35:17 There’s a lot that’s happened last few months with the moratorium, the action plan. What are some of the
0:35:20 things that we’re excited about right now? What are some of the things we’re, we’re less excited about
0:35:25 right now. Why don’t we give a breakdown of where we’re at right now? So I think given what
0:35:28 Collins described about where things were a couple of years ago, it’s great to see
0:35:33 the federal government, um, certainly the executive branch, but not just the executive branch. I think
0:35:39 this is in Congress across both aisles being supportive of frameworks that we think are much better
0:35:46 for little tech. So trying to identify areas where regulatory burden outweighs value and where we can
0:35:51 right-size regulation to make it easier for AI startups. As Colin said, support for open source,
0:35:54 we were in a really different place on that a couple of years ago. Now it seems like there’s much more
0:35:58 consensus. And again, it actually was across the end of the last administration and the current
0:36:04 administration around the value of open source for competition and innovation. Um, there are the,
0:36:10 the national AI action plan also had, um, great stuff in it about, um, thinking through the balance
0:36:13 between the federal government and state governments, which is something that we’ve done a lot of
0:36:17 thinking about. There’s an important role for each. Um, but the, we think the federal government
0:36:22 should really lead regulation of development of AI states should police harmful conduct within
0:36:26 their borders. And I think there’s stuff in the action plan that would try to ensure those respective
0:36:31 roles. There’s also a lot of stuff in the action plan that wasn’t really talked about much. It wasn’t
0:36:37 sort of the headline grabbing stuff that I thought was incredibly compelling, um, in terms of, again,
0:36:42 trying to, to create a future for AI that just works better for more people. And a really good
0:36:47 example is the stuff on worker retraining, um, that focused on different programs that
0:36:53 could help workers if they’re displaced as a result of AI, as well as monitoring AI markets and labor
0:36:58 markets to make sure that we understand when there are significant labor disruptions. So I think it sort
0:37:02 of gets at a point that you were alluding to a couple of minutes ago about like, what happens when
0:37:06 there’s something really disruptive in the future? Can you predict with certainty that there won’t be
0:37:13 this crazy disruptive thing? And no, we can’t. There, there might be significant labor disruption. Um,
0:37:17 others at the firm have talked extensively about how typically there’s always, there are worries about
0:37:20 labor disruptions when there’s new technology introduced. Typically there are increases in
0:37:25 productivity that end up being good for labor overall. We think that’s the direction of travel, but you
0:37:30 never know. We can’t predict it with certainty. And so I think it’s a really strong step to try to just
0:37:35 monitor labor markets to see what the disruption might look like so that we’re set up to take strong policy
0:37:40 action in the future. Can I, can I just say one thing about the AI action plan? Sure. And I, I don’t
0:37:47 want to juxtapose this, uh, to what we saw under the Biden administration, which is incredible amount of
0:37:52 activity in the Biden administration, an incredible amount of activity under the Trump administration. But,
0:37:57 you know, look, I, I kind of view these executive orders and these plans that come out from
0:38:02 administration are very, very important. And some of them have true policy. They direct the
0:38:07 agencies to do things to come out with rewards and then take under rulemakings and things like that.
0:38:15 But from an AI action plan perspective, for me, it was so significant because I think it turned the
0:38:22 conversation on its head before it was, we have to, we have to only focus on safety with a splash of
0:38:28 innovation. Yeah. And now it is, we understand how important this is from a national security
0:38:33 perspective. We understand how important this is from an economic perspective. We need to make sure
0:38:39 that we win while people, while keeping people safe. Yeah. Right. And that dynamic and that shift
0:38:44 of rhetoric is incredibly important because what that does is it signals to the rest of the world,
0:38:48 it signals to other governments that this is the position of the United States and will be the position
0:38:54 for the next three and a half years. And this is the position of the United States to the Congress.
0:38:59 So when the Congress is looking at potentially taking up pieces of legislation or taking actions
0:39:04 or even committee hearings, which, you know, for the broad base of what we’re talking about are fairly
0:39:11 insignificant, all of that is sort of kept in mind. So now the conversation has shifted significantly and
0:39:18 that that is really, really important. Speaking of winning, Colin, I’m curious for our thoughts on AI
0:39:23 policy vis-a-vis China, whether it’s export controls or any other, you know, issues we care about.
0:39:27 Yeah. I mean, well, I mean, look, first and foremost, we’ve talked about it already. I mean,
0:39:33 we have to win. Right. And, and I think, I think that that is, that is at, that is at the main thrust of
0:39:39 a lot of what we’re doing here and a lot of the way that we think about this from a firm perspective.
0:39:46 You know, I think first is making sure that the, the founders and the builders can build appropriately
0:39:52 with appropriate safeguards and an appropriate regulatory structure. The second is how do we win and make
0:39:56 sure that America is the place where AI is, is probably the most functional and foundational
0:40:03 vis-a-vis China. Um, you know, I, I, I think that, um, there has been a long conversation,
0:40:07 the diffusion rule that came out from the Biden administration, specifically on export controls.
0:40:17 Um, many, I think panned that proposal. I think that that was, um, a lot of people suggested it was
0:40:22 probably too restrictive. It wasn’t the right way to think about things. I think, you know, we have spent
0:40:29 most of our time, Matt leading this effort has spent most of his time, our time specifically focused on
0:40:35 how are we regulating the underlying models and how are we regulating hopefully the use of these models
0:40:42 versus specifically sort of on the export control piece. What I will say though is very concerning
0:40:47 sort of some of the proposals that came out, um, from the Biden administration, some of the proposals that
0:40:50 we’ve seen in the state level and some of the proposals that we’ve seen and at the congressional
0:40:58 level of federal standpoint that dealt with specifically export controls on models themselves.
0:40:58 Yeah.
0:41:05 And we’re still kind of having this conversation. There’s, there’s a, um, there is a policy set
0:41:08 that has been kicked around for a while. It’s called the outbound investment policy, which is basically
0:41:15 how much U.S. money from the private sector is flowing into Chinese companies. And I, very noble,
0:41:21 laudable, you know, super supportive of that concept. You know, we are a very sort of primary
0:41:26 America, America first sort of organization here. We’re investing primarily in American companies and American
0:41:34 founders. Um, so, you know, we’re, we’re very supportive of it. But when you, when you sort of
0:41:42 edge into the idea that we might inadvertently ban U.S. open source models from being able to be
0:41:49 exported across the country, like by definition of open source, there is no, there are no walls around
0:41:54 these types of things. So that’s one of the areas that we’ve been very, very focused on. Um, and I think,
0:42:00 uh, I think obviously very important to make sure that we don’t have these very powerful technologies,
0:42:06 U.S. main technologies in, in the hands of our Chinese counterparts and the PLA and the CCP using
0:42:12 this against us. But I also think that we need to make sure that we’re not extending too far
0:42:18 and limiting the power of open source technologies to be able to kind of be the platform around the world.
0:42:24 You know, the final point that I’d make here is we do all ultimately and fundamentally have a decision to
0:42:31 make as, as you know, the U.S., which is, do we want people using U.S. products across the world,
0:42:35 which helps for a whole bunch of different reasons, but certainly on soft power from national security
0:42:40 respective? Or do we want people to use Chinese products? The more that we lock down, obviously,
0:42:46 American products, the more Chinese, the Chinese will enter those markets and sort of take a land grab
0:42:46 in that space.
0:42:51 When you get into more what happened with the moratorium and the fallout that ensued?
0:42:56 I think this one is a bit complicated. There was a perception about the moratorium when it came out
0:43:02 that it would have prohibited all state law from, from existing for a 10 year window. Obviously,
0:43:06 that’s a long period of time. I’m not sure we would necessarily completely agree with that policy stance.
0:43:12 That, from our point of view, is a misinterpretation for a whole bunch of different reasons of actually
0:43:18 what the language said. But, you know, sometimes in D.C., a lot of times in D.C., perception is reality.
0:43:26 And that kind of took hold. But I also think that, you know, there are also, you know, strong competing
0:43:33 forces like we’ve discussed, right, from the, I think, the Doomer crowd or the safety crowd that were
0:43:39 very, very anti, that had used all of their tentacles that they’ve spread out over the last decade to try
0:43:45 and move in and try and kill this. I think they also were successful in leveraging some other industries
0:43:49 to try and come in and also move forward to try and kill this thing.
0:43:55 And look, you know, by virtue of the vehicle, the underlying procedural vehicle, this reconciliation
0:44:01 package that it was moving in, it was a partisan exercise. It was going to be Republicans on Democrats,
0:44:08 and that was that, right? And there was nothing, even a prominent A.I. policy that was going to be
0:44:13 dropped in a reconciliation package that was ever going to drag Democrat votes over it because it was such
0:44:18 a big sort of Christmas tree style thing that had all kinds of all kinds of tax reform positions,
0:44:26 et cetera. And if you were in one of those situations, the margins on the votes become
0:44:33 very, very, very small. So all it took was, you know, one or two Republican senators hitching their
0:44:40 wagon to some of these ideas that were out there to tank this thing, right? And look, I think that’s
0:44:46 going to be a situation that you’re going to fight in any sort of political policy legislative
0:44:50 outcome or any sort of any any sort of issue that you’re going to be running within the Congress.
0:44:58 Right. But I think more so than anything, and we heard this repeatedly from a whole bunch of different
0:45:02 people, and this is what we’ve also experienced. The industry was just not organized well enough.
0:45:06 Right. And that’s not just the industry. It’s also the people who care about this thing that aren’t
0:45:12 actually industry stakeholders. The stakeholders who were pro some level of moratorium or some level
0:45:19 of preemption were just not organized. And I think that that was, you know, both an eye-opening moment,
0:45:25 but also an important moment, because I think what we have done in the preceding,
0:45:31 you know, three, four months since this thing has gone down is we’ve taken a long, hard look at what
0:45:37 we need to do collectively from a coalition to be able to be in a better position next time we’re there.
0:45:41 And so what does that look like? Right. I mean, first and foremost, it comes with
0:45:46 writing, doing podcasts, talking about these things, talking about the details of what’s
0:45:53 actually in these proposals and what it actually means for states and the federal government to make
0:45:57 sure that we’re fighting through the FUD that’s coming through because it’s always going to be there.
0:46:03 There’s misrepresentation all over the all over the field. The second piece is let’s all get on the
0:46:07 same page, which I think we’ve, we’ve worked very hard to do and where we can find alignment. We,
0:46:13 I think we’ve found that alignment between big, medium and little. And then I think the third and
0:46:18 probably the most important is what are we doing on sort of the political advocacy side to make sure
0:46:25 that we have the appropriate tools to be able to push forward in a way that ensures that America
0:46:30 continues to lead and that we don’t lose out on this race to China. And that’s, you know, part of the
0:46:36 reason that we have recently announced our donation to leading the future pack, which will have, you
0:46:41 know, several different entities underneath it, which I think is, is designed to sort of be that political
0:46:47 center of gravity in the space. Um, and that will fight at the federal level and the state and local
0:46:52 levels. So we’re happy to be a part of it. And I, we expect, you know, there will be others that join
0:46:58 this sort of common cause fight on the AI side. If we could wave a wand, what would we like to be done at
0:47:03 the state level versus the federal level versus how should we think about that, that interplay
0:47:08 compared to where we’re at now? Yeah. So, so I, I think there, the, the, the helpful answer here
0:47:12 comes from the constitution. Constitution actually lays out a role for the federal government and a
0:47:18 role for state governments. Federal government takes the lead in interstate commerce. So governing a
0:47:23 national AI market and governing AI development, we think is primarily Congress’s role.
0:47:29 Um, sometimes when, when people say that, I think the, what other people hear for some reason is
0:47:35 states should do nothing. And we have been, we’ve tried very hard to be very deliberate in not saying
0:47:40 that and making clear that states have an incredibly important role to play in policing harmful conduct
0:47:45 within their jurisdictions. So criminal law is a perfect example. There is some criminal law at
0:47:49 the federal level, but the bulk of criminal laws at the state level, like when you think about routine
0:47:55 crimes, if you were going to prosecute someone, uh, prosecute a perpetrator, the it’s likely that
0:47:59 that would occur under state law. And so to the extent we want to take account of
0:48:06 local activity that, um, that would, where there’s criminal conduct involved and we want to make
0:48:09 sure that the laws are robust enough to protect people from that activity, that’s going to be
0:48:17 primarily state, state law. Um, oddly enough, I mean, as Colin is describing, like we, this isn’t the
0:48:21 delineation that we’ve started out with. There are a lot of state laws that have sort of taken the approach
0:48:27 of some, sometimes explicitly, um, Congress hasn’t acted. So we have a responsibility to act. And that’s
0:48:32 true to some extent, like you can act within states can act within their constitutional lane.
0:48:37 Some of what states have done have gone outside that lane. And so we actually just this week
0:48:43 released a post on, um, potential dormant commerce clause concerns associated with state laws. And the
0:48:49 basic idea there is that there’s a constitutional test that says that states cannot excessively burden
0:48:55 out-of-state commerce if it, when it, when that greatly exceeds the in-state local benefits.
0:49:00 And so courts actually weigh that there’s a balancing test. Are the harms cost to out-of-state
0:49:06 activity, do those significantly outweigh the benefits on the local side? And we think that at
0:49:10 least for some of the proposals that have been introduced, it’s, it’s likely that they won’t,
0:49:14 that the benefits are somewhat diminished relative to what the proponents think they are,
0:49:19 and that the costs are significant. Like the cost of a developer in Washington state
0:49:25 for complying with a law that’s in California or a law that’s in New York is going to be significant.
0:49:30 And so our hope I think is not that the dormant commerce clause ends up serving as a function that
0:49:37 makes it hard for states to enact laws. Um, but actually just get serves as a guidepost for states
0:49:41 around the kinds of laws that they might actually introduce. And I think it pushes in the direction
0:49:46 that’s consistent with our agenda, which is to, to take an active role in legislating and
0:49:52 enforcing laws that are focused on harmful use. Looking at the next six months to a year,
0:49:55 what are the issues that we’re most focused on or that we’re thinking about are going to,
0:50:00 you know, be playing a role in the conversation? Yeah, I think it’s first and foremost,
0:50:05 some level of federal preemption. And I, I, I want to be very specific about this. Again,
0:50:10 to Matt’s point, we’re not talking about preempting all state law. We’re talking about
0:50:16 making sure that we have a federal framework specifically for this model regulation and,
0:50:23 and hopefully how the models can be used. Right. Um, I, I, I think that’s going to be so,
0:50:30 so critical because we can’t just like any other technology, no technology can live under a 50 state
0:50:35 patchwork. And, and, and that’s, that’s been the biggest issue that we’ve been fighting over the
0:50:43 last year and a half or so. Um, so I think, I think that, um, I think that there are some other
0:50:49 sort of policy sets that I think will be handled beyond that, that I think can kick into sort of
0:50:54 workforce training. I think there’s some literacy things that should be coming up. Obviously,
0:50:59 there’s a huge, robust conversation around data centers and energy that I think it will be really,
0:51:04 really important. But above all, I think most of our time and energy will be focused on trying to
0:51:09 have some level of federal standard here to try and drive the dividing line between the federal and
0:51:12 state government, which I think Matt has already done a ton of great work on.
0:51:19 Yeah, I think this is just a super exciting policy moment for AI. Um, there’s the, the last couple
0:51:22 years where I think there are a bunch of ideas that have been proposed. And for the reasons that we’ve
0:51:28 discussed, we think those ideas fall short, both in terms of protecting consumers and in terms of ensuring
0:51:34 that there’s a robust startup ecosystem. Um, most of those laws I think have actually not succeeded in
0:51:39 passing. So like there were a number of laws introduced at the state level in the, in this past year’s
0:51:44 legislative sessions that we thought had a strong likelihood of, of passing. And I think to date,
0:51:49 none of them have passed. Um, Colin has also been building out the expertise and skillset and
0:51:54 capacity on his team. We just hired Kevin McKinley to lead our work in state policy. And he, I think,
0:52:00 will help us to take a real affirmative position in the legislative sessions ahead on what might actually
0:52:04 be AI policy. That’s good for startups. So instead of being in the position of saying no,
0:52:07 because we’re sort of starting late and kind of with one hand behind our back,
0:52:13 I think we’re in a position to really actually try to articulate and advance a proactive agenda
0:52:19 in AI that’s compelling. I think Colin hit the main parts of it. Um, ensuring proper roles for the
0:52:24 federal and state governments, focusing on regulating harmful use, not development. And there are specific
0:52:29 things that you can do there in terms of increasing capacity and enforcement agencies, making clear that
0:52:35 AI is not a defense to claims brought on under existing criminal or civil law. Um, and in technical
0:52:40 training, I think for government officials to make sure that they can identify and prosecute cases where
0:52:44 AI is used in a harmful way. And then all this infrastructure and talent stuff that, that Colin’s
0:52:51 describing work or retraining AI literacy. Um, we’ve also given some thought to the idea that has been
0:52:57 articulated by a number of lawmakers and was in the national AI action plan of creating a central resource
0:53:00 housed in the federal government. And you could also do it in state governments as well.
0:53:05 Uh, that lower some of the barriers to entry for startups. Um, you know, compute costs and,
0:53:11 and data access. Um, and we think that’s really compelling in terms of, um, ensuring that startups
0:53:15 can compete. And that idea, like many of these is bipartisan. It’s been supported by the current
0:53:20 administration. It was supported by leading Democrats, um, over the last couple of years. So that’s the kind
0:53:26 of thing that we are hoping that when we have the, the room and position to really advocate for an
0:53:29 affirmative agenda that we’ll get some traction in policy circles.
0:53:36 We’re not always in 100% alignment with other people in the industry, you know? And, and I think,
0:53:41 I think that that’s, you know, big, medium, little, you know, across the board, there’s other sort of
0:53:46 like consumer advocacy groups that obviously feel differently about these things. I think for the
0:53:52 most part, the industry is generally aligned on some level of a federal standard here and understanding
0:53:57 that the thing again, that won’t work is a 50 state patchwork. Yeah. And I think that that’s
0:54:02 super, super important because I think for the first time you actually have this sort of alignment
0:54:06 there. And if you have that sort of alignment, that’s kind of momentum that you can to actually
0:54:10 push things over the finish line and get something done. And I, and I think, look, also the Trump
0:54:15 administration to their credit has also been incredibly supportive of this idea too. There’s a,
0:54:21 like, that’s an incredibly important point. One criticism usually raised in sort of an
0:54:26 implicit criticism sort of way is, Hey, you’re the little guys, but often you align with the big
0:54:31 guys. So aren’t you just saying, aren’t you just in favor of a deregulatory agenda that works for big
0:54:35 tech? And one of the things that I think is really extraordinary about the little tech agenda is it’s
0:54:41 really nonpartisan and it’s doesn’t take a position on big, little, it basically says, here’s the agenda.
0:54:46 And when you agree with us, we’ll support you. And when you disagree with us, we’ll oppose you.
0:54:51 And that’s not party line. It’s not big, little. And so I think what we saw over a
0:54:58 the phase that Colin was referring to kind of initially in the recent set of AI policy was a
0:55:04 phase of divergence between big and little licensing regime. Bigs were sort of pushing it. Little was
0:55:08 concerned about it. Then there’s, then there was a period of convergence. And I think actually,
0:55:13 if you look at like the national AI action plan comments across a range of different providers,
0:55:18 as Colin’s saying, like a lot of them, they had some core similarities. So, so lots of large companies
0:55:23 have advocated for federal preemption. We don’t oppose that just because big companies are advocating for
0:55:27 it. We think that that’s good for startups. I think it’s possible. I’m curious. I mean,
0:55:33 this is really, you know, Colin really understands this in a way that I don’t like how the political
0:55:37 chips will fall. I think it’s possible we’re in a period of some divergence. And one thing that we hear
0:55:42 repeatedly, which is sort of funny, is people will bring us stuff and they’ll say industry agrees
0:55:46 with this. So we expect you to agree. You can’t, the industries already agree. You can’t disagree.
0:55:51 And we say the big parts of the industry have agreed, but we, we, sometimes we agree with them,
0:55:55 but sometimes we have different views. And so when we disagree, it’s not because we’re trying to like
0:55:59 blow up a policy process or make it different difficult for lawmakers who are trying to move
0:56:03 something forward. It’s because when we’re looking at it, we’re looking at it through this particular
0:56:08 lens. And I think, I hope it’s not the case, but I think there might be more fracturing in the
0:56:11 months ahead. Yeah, I agree with you on that. And by people, he means lawmakers,
0:56:16 just to be specific. Yes. That’s a great place to wrap. Colin, Matt,
0:56:18 thanks so much for coming on the podcast. Thanks very much.
0:56:26 Thanks for listening to the A16Z podcast. If you enjoyed the episode, let us know by leaving a review
0:56:32 at ratethispodcast.com slash A16Z. We’ve got more great conversations coming your way. See you next time.
0:56:37 As a reminder, the content here is for informational purposes only.
0:56:52 Please note that A16Z and its affiliates may also maintain investments in the companies discussed in this podcast.
0:57:00 For more details, including a link to our investments, please see A16Z.com forward slash disclosures.
0:57:08 Thank you.