Pod Insights Equity Podcast California just drew the blueprint for AI safety regulation with SB 53 About this episode California just made history as the first state to require AI safety transparency from the biggest labs in the industry. Governor Newsom signed SB 53 into law this week, mandating that AI giants like OpenAI and Anthropic disclose, and stick to, their safety protocols. The decision is already sparking debate about whether other states will follow suit.
Adam Billen, vice president of public policy at Encode AI, joined Equity to break down what this new law actually means and why it managed to pass its predecessor SB 1047 incurred so much ire from tech companies that Newsom ended up vetoing it last year.
Listen to the full episode to hear about:
What "transparency without liability" means in practice, and whether it's enough to ensure safe AI is released to the masses.
Whistleblower protections and critical safety incident reporting requirements.
What's still on Newsom's desk, including regulation on AI companion chatbots.
Why SB 53 is an example of light-touch state policy that doesn't hinder AI progress.
The battle for federalism amid moves to take away states’ rights to enact AI regulation.
Equity will be back Friday with our weekly news roundup, so don’t miss it.
Equity is TechCrunch’s flagship podcast, produced by Theresa Loconsolo, and posts every Wednesday and Friday.
Subscribe to us on Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.
Learn more about your ad choices. Visit megaphone.fm/adchoices
Transcript 0:00 Speaker 00 This episode of Equity is brought to you by Schneider Electric. Schneider Electric, your energy technology partner.
0:05 Speaker 01 Hello and welcome back to Equity Tech Crunch's flagship podcast about the business of startups. I'm Rebecca Bollan and this is the episode where we bring on industry experts to help us explore a trend in the tech world and dive deep. California just became the first state in the nation to require AI safety transparency from the biggest labs like OpenAI and Anthropic after Governor Newsom signed into law SB 53. And it's already sparking debate about whether other states will follow.
0:37 Speaker 01 To walk us through it, today we're joined by Adam Bilan, Vice President of Public Policy at youth-led advocacy group ENCODE AI. Adam, welcome to the show.
0:45 Speaker 02 Thanks so much, Rebecca. It's great to be here.
0:48 Speaker 01 So tell me a little bit about ENCODE AI and what your mission is and when you were founded and what kind of things you're working on today.
0:56 Speaker 02 Yeah, so we've been around for about five years. We were originally founded by our president, Sneha, in California. And for many years, we're doing sort of public education workshops. We worked quite extensively in a couple of states on facial recognition technology in schools, among other issues. And then over the last two years or so, have sort of tilted very directly into advocacy. And so the first big bill we worked on was SB 1047, which was the AI safety bill in California last year that was vetoed by Governor Newsom. But we also do work in Congress and our sort of, I think, unique thing that we try to do that a lot of other groups don't is we try to work across sort of the whole range of risks from AI. So we don't just work on AI safety legislation. We also work on child safety legislation. We work on facial recognition technology.
1:41 Speaker 02 We work on deep fake porn. We worked with Senator Cruz's team on the Take It Down Act in Congress. We were spending a lot of time pushing back against the AI moratorium recently. And we try to sort of bring together coalitions and groups of advocates that normally don't work together to try to get things done that maybe we wouldn't be able to accomplish siloed off in our little miniature coalitions. And I would point to the win on the moratorium as the first time that I think every single one of those little mini tech coalitions was unified on a single thing and successfully beat something that we did not expect, honestly, to win on and thought at many moments during that fight that we would lose.
2:20 Speaker 01 Well, okay, so for those who don't remember, the AI moratorium was a proposal that was going to be part of the budget bill that would have banned states from regulating AI for a period of 10 years, which, you know, sounds, yeah, it seemed odd. And no other industry, I think, has been given the potential for such an allowance of a lack of oversight, considering there's no federal regulation. So that was struck down, but Senator Cruz actually is working to bring up something similar like it. And you've got, you know, the whole of Silicon Valley's hundreds of millions put into super PACs to, you know, fight for candidates that have a light touch or a no-touch approach to AI regulation.
3:05 Speaker 01 Yeah, thoughts on that coming back, that you're gonna have to pick that sword up again as a little pacifian?
3:09 Speaker 02 Yeah, no, I, you know, we've been running a coalition of, I think the calendar invite has 320 people on it now from, you know, 200 or so different organizations all that were working to beat the AI moratorium when that was in Congress. And we're all sort of gearing up for what this next fight is going to look like. And the thing we're all sort of expecting is that they're going to come back and say, you know, last time that moratorium, you know, that was unreasonable. We didn't really want that. What we really want is, you know, a federal standard that preempts the states. We're actually going to do a bill now. The thing I fear is that they will pass a federal standard maybe on one particular issue. Maybe they'll authorize the AI Safety Institute, which is something we support, although pass whistleblower protections, which we also support. But they'll do that,
3:53 Speaker 02 and then they'll preempt a much broader set of state laws and sort of pitch that as a reasonable compromise with the people that were originally opposed to the moratorium. But the conversations we have been having sort of coalitionally is like, well, if we only get a win on one issue and then Congress is done acting on AI, but we lose the ability of the states to do anything else, then we've deleted federalism for the most important technology of our time, and we've gotten a federal standard on maybe one thing. And it remains to be seen whether that federal standard will even be good enough. And, you know, SB 53 in California, which just passed, is our bill. We've been working on that bill and 1047 last year for two years now. I love that bill. I think it's a very good bill.
4:33 Speaker 02 If you told me that SB 53 was the bill that would replace all the state bills on everything related to AI and all of the potential risks, I would tell you that's probably not a very good idea and that this is a bill sort of designed for one particular subset of things. And I think you could make the same argument for bills on child safety or data privacy or training data transparency, et cetera. So that's the sort of next fight that we're gearing up for.
4:56 Speaker 01 Yeah, I mean, it is like the American way is to have, not everything can be done via federal regulation. It just moves too slow. States often can lead the way when it comes to regulating industries and then kind of setting a standard that other states might follow, right? So California often leads in terms of environmentalism, for example, and now it's leading in terms of AI. So, you know, you mentioned SB 1047. This was a similar type of AI safety bill that was introduced last year and then also vetoed last year. I can't remember exactly when it was introduced, but by the same author, Scott Weiner, Senator Scott Weiner. And so now SB 53 has passed and been signed into law by Governor Gavin Newsom. I believe there's a similar legislation that Governor Kathy Hochul in New York
5:41 Speaker 01 is gearing up to hopefully sign or possibly veto. Can you walk us through what SB 53 actually means? And then after, I'm going to ask you how that's different from 1047.
5:51 Speaker 02 Yeah, absolutely. So SB 53 is really a transparency focused bill. When Governor Newsom vetoed SB 1047, he started up this working group that studied what the state should do about AI. And they came back with a bunch of recommendations. They were also tasked with identifying whether the types of risks we were trying to address in 1047 last year were real and worth addressing. And they came back and said, one, yes, these risks are real. And two, here are some things you should do about that. So the sort of core of the bill is these frontier policy frameworks formerly called safety and security protocols. And it's basically the idea that a company has to have a plan for how they're making their models safe to prevent catastrophic risks, things like their models helping someone commit cyber attacks
6:35 Speaker 02 on critical infrastructure or create a bioweapon, things that many of them are. Already doing internal evaluations on and have publicly committed that they'll do both to the previous presidential administration and at the International AI Safety Summit in South Korea. So it says you have to have that plan, you have to publish that plan publicly, and then you have to follow that plan. So it doesn't say you need to do XYZ specific things in your safety and testing processes. It just says you need to tell us what your plan is and then follow it. On top of that, it has incident reporting. So if a dangerous incident happens due to a company's models, they have to report that to the Office of Emergency Services in California. And then there are also whistleblower protections for employees in the state so that they can come forward to the government and say this is
7:19 Speaker 02 something dangerous that I think is happening at my company, which we've seen happen, you know, repeatedly in social media, for example, with people like Francis Haugen and Archer Obehar. And then finally, it has a provision called Cal Compute, which has been the sort of darling of the bill that has never been criticized by basically anyone, which is the idea that California should have computing resources for academics that might not have access to the chips and GPU and compute power that it really does take to do sort of cutting edge AI research. And right now, all of that is compute is concentrated in the biggest companies and some small or medium sized companies and startups. But this says, you know, California academics should also have have access to those resources. Interesting. How do they get the access to those resources? Like, who gives it to them?
8:03 Speaker 02 Yeah, so it'll be housed under a California university, but California somewhat famously right now is in a severe budget deficit. So the bill doesn't actually assign any money to Cal Compute yet. So we'll have to come back and actually appropriate money for them to build the physical computing structure and then spin up the expertise to properly distribute the resources. But this isn't unprecedented. In New York, there's Empire AI, which is a similar process where they built up a bunch of compute and then distributed out to academics in the state.
8:32 Speaker 01 Yeah, that is really important. You know, I spoke to Karen Howe, the author of Empire of AI, a few weeks ago on the podcast, and she talked about how academics are really being kind of pushed out of the conversation around what AI could look like, what new models we could possibly promote. She said that the vast data scaling that we're doing, the scaling of infrastructure, is the intellectually lazy thing to do, where a lot of academics think there might be actually a more efficient way to do this. It might just take a little more time. But, you know, there's this race to get to the top. There is no time, which I'm sure you're seeing quite a lot of yourself, or at least that kind of dialogue. Okay, so this is what this bill does. What about how it's different from 1047? Because Silicon Valley was up in arms about 1047.
9:17 Speaker 01 They were absolutely freaking out last year that it was way too much. And this year, not so much. That, well, I mean, there was some opposition, but not as loud.
9:24 Speaker 02 Yeah, it was very quiet and soft and anthropic indoors. I think Google was publicly neutral, and Meta even made some kind of mildly positive comments in the news at some point. The key thing that's different about this bill is that SB 1047 was a bill partially about liability and figuring out when and how companies should be liable for their systems. And I think this is maybe the most misunderstood thing about that bill. A big part of our intention in writing that bill was that right now there is an existing tort liability system, and you can sue an AI company for harms from their products. This is happening right now. All of the children that have died by suicide, where there are lawsuits, that is under the existing tort system. You don't need to pass a law to say that AI developers can be sued. That's not how
10:05 Speaker 02 the American legal system works. The goal of SB 1047 was, what does reasonable care look like for a developer to make their system safe before they release them? And that cuts both ways. If you're a developer and you're a general counsel for that company right now, and you're trying to figure out, how do we make sure that if we're sued, we can show the judge and the court that we've been doing things to make sure our products are safe? You have no idea what that looks like, because there's no precedent, and there's no law that tells you what to do. It's all relying on sort of people's guesses as to what courts will think. So the bill said, here is one thing that you can do that would look like exercising reasonable care. Which was the safety and security protocols. That had people up in arms because I think it was read as we're going to sue AI developers into the ground or something.
10:50 Speaker 02 But really, in many ways, it was a liability shield as much as it was exposure to new liability. If you didn't have any safety and security protocols, or you published your plan and your plan was that we'll do nothing, and then you were sued. Sure, probably you're exposed to more liability. But in the reverse, if you had a plan and you published it, then that could help protect you from liability in those circumstances. So the key difference with SB 53 is that it does not have that same language around liability in it. It's pure AG enforcement, and it's all about just transparency about publishing the plans. There are a couple other, many other little technical things around the edges, but that's really the core difference between the two bills.
11:26 Speaker 01 Okay. Well, there is some liability in another bill that Newsom's probably weighing up called SB 243 that I want to get to. Before I do, I wanted to just double click on SB 53, the reporting requirements for safety incidents and things like that. It seems, I mean, I used to cover transportation, a lot of autonomous vehicle work, and with AI that goes out into the world and can physically harm people, there is reporting requirements, right? Companies need to report if they have any incidents, they have to report how many disengagements they've had. There's not, there's no standard for having some kind of government oversight as to the safety of a technology that's being deployed en masse.
12:09 Speaker 01 So I just wanted to hear a little bit more about your thoughts on the reporting requirements and what they actually mean.
12:13 Speaker 00 What moves our world forward? Energy. History has been shaped by it, revolutions won with it, and progress dependent on it. For 190 years, Schneider Electric has been a leading force in this story, creating the technologies that harness energy and transform the competitive landscape. And today, in the dawning era of energy and industrial intelligence, Schneider is helping customers build more efficient, resilient, and sustainable operations across every sector. From data centers to homes, from buildings to industry. Schneider Electric, your energy technology partner. Discover more at se.com.
12:50 Speaker 02 Yeah, for sure. I mean, I think that's totally true for certain types of technology. And this sort of gets back to the root of SB 1047 and SB 53, which is that we don't have frameworks to think about these systems that are not being developed and deployed in the same way that more narrow types of AR are being deployed, for example, in autonomous vehicles. So there's no framework for a frontier model developer right now to disclose if they've had an incident where, I think like Anthropic reported recently, that people were using gigantic versions of their systems to go and commit fraud, I think in places like North Korea. There's no existing system where Anthropic has a place, even if they wanted to, to report that incident to the government. And so this is just saying, like,
13:32 Speaker 02 just as this is true, as you're saying, in other fields, we should have some equivalent system for these sort of catastrophic types of harms that up until now, there hasn't been a system for. And one thing we do in the bill is say, the Office of Emergency Services, which is responsible for handling these reports, if a federal standard comes into effect through Congress or in somewhere like NIST or the AI Safety Institute or elsewhere, probably somewhere with rulemaking powers like the Department of Energy, OES can actually designate that standard as good enough for compliance with the California bill. And so part of what we're saying is like, it seems like the government federally will do something like this at some point. And there's been some indication in this admin and the last
14:14 Speaker 02 admin that that will eventually happen. But until that point, this is kind of the best we can do. So this is also kind of in the context of this whole preemption conversation. I think people are understandably wary about doing things that feel national security adjacent or focused at the state level. Part of what we're saying is, that's probably reasonable, but you don't need to preempt all the state bills. Here's a way where if you pass any standard federally without preemption, we will automatically choose to give those powers to the federal government.
14:43 Speaker 01 Okay, that makes sense. And yeah, it's interesting what you say about, you know, potential cyber attacks because we had Ami Lutwak, who's the CTO of Wiz on recently as well. And he spoke a lot about how people are using AI prompt based attacks, vibe coding in malware. It's a growing field and these things will be happening a lot. So it is really, you know, important to track them. But then, you know, going back to the liability piece, right? I'm not sure if you have any insights into what's going on on Governor Newsom's desk. But I covered, you know, a few weeks ago, that SB 243 passed the California State Legislature and that's regulating AI companion chatbots. It's requiring safety protocols and it does create some
15:27 Speaker 01 liability for AI companies handing out these companions that don't follow certain safety protocols or if there's harm. Did you work on that bill as well?
15:37 Speaker 02 We didn't. We've been supporting AB 1064, which is another bill focused on AI companions. 243 is the Padilla bill and 1064 is the Rebecca Bauer Cahan bill. And so we've been focused on that one, which has some more specific requirements around sort of types of things that a chatbot should or should not talk with a minor about. And so I've been tracking 243, but we've been more focused on that bill.
15:59 Speaker 01 And so this is in California as well, or?
16:01 Speaker 02 This is in California and has also passed and is on the governor's desk. Common Sense Media is the primary sponsor on that bill. And so we did a lobby day with them in California on that bill and SB 53 as well. The conversation around AI companions is very interesting. And two years ago, I think we were some of the first people sort of yelling into the void that AI chatbots supplanting human relationships probably is bad and not something that we should let happen. And at that time, it was like everyone thought we were crazy and that this was not a real problem. And now it feels like it's like the biggest national conversation around this technology. And I think the way we're viewing it is that there are almost like three sort of buckets of types of AI companions. There's sort of the worst of the worst. I'm thinking about Replica, Character.ai, the meta companions
16:46 Speaker 02 that were rolled out that sort of explicitly engage in sexual content and conversations with minors.
16:51 Speaker 01 Technically, it was romantic and sensual, but yes.
16:55 Speaker 02 That was the internal policy, although I would definitely.
16:57 Speaker 01 Oh, yeah. Well, we've seen. Yeah, you're right. Yeah, fair enough.
17:01 Speaker 02 Yes, but it is. It is stunning that their internal policy was was sensual, which is I kind of can't believe that their counsel approved that, but whatever. I'm not the first time Meta has done something crazy. And then there's the sort of middle section of bots. I would probably put like ChatGPT Claude, maybe Google's bots like Gemini in this category. That are, I think, are clearly not being designed expressly for this purpose, but where some of these harms are happening anyways. And, you know, this came up with the Adam Rains case, the young boy in California who died by suicide after GPT. Four O gave him sort of meticulous instructions and encouraged him. Where I in no way think that OpenAI went from the ground up and was like, I want our bot to engage.
17:43 Speaker 02 In sexual conversations with minors or instill them in relationships. I really do not think that. And I think that is, that would be a misreading. But I do think that clearly not enough was done in the safety. Steps within their company, and we know that the rollout of specifically GPT four O was rushed into a nine day period. And I would also place this in the broader context of their restructuring from a nonprofit to a for profit. And at the time they were rushing to get GPT four out to race Google's release of their next model. This was, I think, twenty twenty three or twenty twenty four. And so there's this muddier ground with those models in the middle, but I think no matter what. There's some set of features that you definitely don't want these bots doing, and I like that ten sixty four and we like that ten sixty four is sort of more specific about that.
18:25 Speaker 02 Whereas two forty three, I think, is maybe more vague and serious. I think it's still a good bill that would be good if it passed, but is less where I think our our mindset is.
18:35 Speaker 01 Yeah, I think that if either of those bills passed, what it says to me is that for government to dip its toe in the water of fighting against. Big tech companies, it requires wanting to protect children.
18:49 Speaker 02 Yeah.
18:49 Speaker 01 It seems like there's a lot more leeway for regulation when we're talking about harms to children.
18:55 Speaker 02 Yeah. No, I think that's true. And, um, you know, this. It was true in Congress with the Kids Online Safety Act, which which moved very quickly. And I think if that was not a bill about children, it would not have moved in the same way. But I mean, I think this is for very good reason, right? Like, I find it kind of stunning, honestly, that the same social media companies that sort of ran over kids, I think, for basically a decade and moved fast and did, in fact, break things, are the same ones developing AI, right? Like, the frontier model developers are the big social media internet companies. It's like Google, it's Meta, it's XAI. And the most stunning development to me has been that it used to be that you had to go out and download Replica or Character.ai if you wanted to talk to one of these bots.
19:39 Speaker 02 Now that they're being integrated in Instagram, in X, there's no avoiding them. And it's a bit stunning to me that those companies didn't see the same political backlash coming on kids and AI that they have gotten for like five or 10 years on social media. Like, it seems odd to me if you're government affairs at one of these companies and you've been getting slammed and called to testify before Congress on harms to kids, that you would not be very intentional to make your products safe for kids. And so I think part of what is happening and the justifiable reason that kids' issues are such a political touchpoint is that it feels like the companies have just willfully decided to ignore the lesson that we thought maybe they learned from social media, but that clearly they haven't.
20:24 Speaker 02 And they also successfully did kill social media legislation in Congress and in many states, or have sued it out of existence. And so they, I think, see clearly that government has failed to regulate on social media and they can run the same playbook again. And you mentioned the super PACs earlier. This is the exact same strategy that the crypto world successfully used to push their fair shake super PAC, which I think had $150 million or something. And it is literally the, I'm guessing, the same strategy, but probably on AI with even more money and with a more diverse set of super PACs. You have Leading the Future, the OpenAI, A16Z super PAC, you have the Meta super PAC, I think NetChoice has announced a super PAC. And you also see them, and I think this is new in comparison to crypto, have a very
21:08 Speaker 02 specific focus on state legislation. Meta has a super PAC just for California that will have tens of millions of dollars, according to them. Leading the Future has named a number of US states as targets for their dollar spend, hard dollar spend. And that is that is unique from previous political fights. And I think they see this chance to repeat the social media political fight. Plus the hard dollar spend of the crypto fight. And I think they also see the potential for kids' issues to hurt them on this. And I think part of the reason you didn't see a super PAC on social media kids' issues at the scale that you're seeing for crypto and AI is that they see that as a vulnerability for them. And if they come out with the $200 million super PAC to kill bills like the Kids Online
21:49 Speaker 02 Safety Act that are on a very narrow band of things, but then basically no one disagrees should be dealt with. It's a much harder thing for them to swallow and it's harder for them to get politicians to accept that money when it feels like blood money that has been spent specifically to kill legislation to protect kids. And I think they're seeing this opportunity with AI.
22:11 Speaker 01 Well, COPA was killed, right?
22:12 Speaker 02 It was killed.
22:13 Speaker 01 Sorry. Yeah, the Kids Online Safety Act, yeah.
22:16 Speaker 02 It was killed and they did it without a super PAC on this scale. So they maybe didn't even need one this time. But I would expect that with this new AI super PAC, they will use the AI issues as cover to some extent. But, you know, God knows that they will also be coming after the social media bills in the states and in Congress under the cover of this AI super PAC.
22:34 Speaker 01 Yeah, I mean, it's funny you say that, like, haven't they learned anything about children's safety and harms to children and how they should avoid it? I think that, you know, they've actually learned, well, we can harm children and it's fine and we could still operate and there is no backlash. There's no real backlash that actually hinders the cost of doing business for us. You know, if they spent half the money that they spend on lobbying on instead child safety, then this probably wouldn't be a problem at all. Broadly, there is this conversation happening in Silicon Valley that any regulation is anathema to progress and innovation and we can't afford to slow down at all. And, you know, Peter Thiel has been likening like any regulation as a means of slowing
23:18 Speaker 01 down AI as a means to the Antichrist, apparently, whatever that means. How do you view this as someone who has to, you know, dance with policymakers?
23:27 Speaker 02 I think the reality is that policymakers themselves know that we have to do something and they know from working on a million other issues that there is a way to pass legislation that genuinely does protect innovation, which I do genuinely care about, while making sure that these products are safe. And I don't think that they buy that this complete binary and bifurcation. Is real and I, you know, I think you've seen this with the conversation around SB 53. When you get to a bill, and I think maybe 53 and 1047 had more scrutiny and eyes and, you know, stakeholder input than maybe any other tech bill in the country. You can actually get to a version of something that the companies are largely okay with,
24:08 Speaker 02 but you have to do it through the long, very hard process of doing democracy in action, which is very ugly and messy. And I think the companies would have you think in Congress that you can snap your fingers and pass a bill in the states on technology. But as someone trying to do it, I can tell you that it is not very easy to do in the states and states are still a process, are still a political process where things can die at many, many points. And so basically, I don't think that policymakers completely buy this. And I think the reality is that if the thing you care about is beating China in the race on AI, and I do care about that, the things you would push for are stuff like export controls in Congress.
24:50 Speaker 02 Like you would make sure that American companies have the chips, but that's not what industry is pushing for. That's not what's happening. No, they're like, oh, well, it's a pay us enough. Yeah, exactly. That's such a good point. And I think some of this is complicated by the fact that, you know, the companies are dependent on the chip manufacturers like Nvidia, and so they don't want to make them mad if they need to get Nvidia to sell a bunch of chips to them. But I think maybe if you, if you asked Dario or Sam Altman or whoever on substance if they think export controls are good, they would, they would probably say yes if you gave them truth serum, but that is not the political context in which we exist. So look, like, does the race with China matter? Yes, absolutely. Should we make sure that American companies win? Yes, absolutely.
25:33 Speaker 02 Should we pursue policies that actually do that, like export controls and domestic investment? In chip manufacturing, yes, absolutely. Are bills like SB 53 the thing that will stop us from beating China? No, I think it is just genuinely intellectually dishonest to say that that is the thing that will stop us in the race. And I think policymakers are seeing that and you see people on the Hill moving towards bills like the Chip Security Act that would put export controls on China. But in the meantime, there's going to continue to be this kind of propping up as that being the narrative to kill state bills that are actually quite light touch. Like SB 53 is a transparency bill. Like I said earlier, if you told me it was the silver bullet solution to every risk from AI, I would tell you, no.
26:17 Speaker 02 And I think the fact that you have companies publicly supporting it speaks to that fact. It's a great bill. It does very important things about risks that I really do see coming down the pipeline in the future. But it's not a silver bullet. It's not a full solution and it's definitely not going to stop the labs from beating China. And I think you can also point to the fact that these companies are already doing the stuff that we ask them to do in the bill. They already have protocols that describe how they test their models. They do do safety testing on their models. They do release model cards. Are they starting to skimp in some areas at some companies? Yes, and that's why bills like that are important. But part of that is that even some of these companies have written into their charters or internal documents, policies that if another lab stops doing safety testing, they will also stop.
26:59 Speaker 02 So even if these companies want to do the right thing, you kind of need to give them the competition policy excuse to be on the right side. And I think that's why you see companies endorsing bills like this and pushing standards like this, because I don't think that people inside these companies are universally bad.
27:18 Speaker 01 Well, they're not bad, but they have a financial incentive and that skews how they're going to perform. And that's where regulation comes in. That's how capitalism works for everyone. So like, you know, you need capitalism to improve innovation and competition and create amazing new things, but you'd also need regulation to come in and have some guardrails to ensure that not only the profit motives of the shareholders at the top are considered when we're talking about such a disruptive technology.
27:46 Speaker 02 Yeah, I mean, I think, to be frank, like. I don't like this broad vilification of the companies building AI. I think that is the incorrect way to view this. They are doing their jobs. They're doing their jobs. They're making incredible technology. And policymakers also have a job in that context that the companies recognize. And that's the reason that they negotiate with policymakers in California and in Congress, and why they come out with their own proposals. And that process of democracy and federalism is the entire foundation of our country and our economic system. And I hope that we will keep doing that successfully. And I think SB 53 is one of the best proof points that that can still, that that can still work.
28:23 Speaker 01 Okay, well, that's a great point to end it on. I really appreciate you coming on and talking to us about this. If any of our listeners want to connect with you online, where can they find you?
28:34 Speaker 02 Yeah, we're at encodeai.org and I think encode action on Twitter. Yeah, and my email is on the website, so feel free to reach out.
28:44 Speaker 01 Okay, great. Well, thanks again for joining us on the show. To our listeners, you can find me on Wiz Guy and X, and you can find Equity at Equity Pod on those same platforms. Talk to you next time.
28:58 Speaker 03 Equity is hosted by TechCrunch senior reporters and produced by Teresa Locan Solo with editing by Kel. Subscribe wherever you get your podcasts and see what's next at techcrunch.com/events. Thanks so much for listening and we'll talk to you next time.
Previous From $100B OpenAI deals to $100K visa fees Next AI slop, government stops, and startup uncertainty Page updated October 12, 2026 · transcript and summary are machine-generated