That UL safety logo is a lot more complicated than it looks

Support for Decoder comes from Adobe. Life is unpredictable, and that means you need your projects to adapt with whatever gets thrown at you. That means mastering the ability to pivot and collaborate with others to reach your goals. Adobe gets that, which is why they made a tool that's just as flexible as you are. PDF spaces and Acrobat. Your PDF files are no longer static. Instead, they're living documents that flex with you and your project's needs. Learn more at adobe.com slash do that with Acrobat. Hello and welcome to Decoder. I'm Eli Patel, editor-in-chief of The Verge, and Decoder is my show about big ideas and other problems. Today, I'm talking to Jennifer Scanlon, the CEO of UL Solutions. That's Underwriters Laboratories. You know, the UL logo listed on all your electronics. That symbol means it's been tested and found safe in a variety of ways. UL has been around for a hundred years. It started as a way for insurance companies to do fire and safety testing on electrical products just as electricity was coming into homes. But now UL is everywhere, and it's one of those companies that we really like to poke at here on Decoder, because it's basically hidden in plain sight. That logo is on everything. But scratch the surface, and the business of UL is pretty complicated. There are tons of cheap, non-UL products on Amazon full of lithium-ion batteries. Maybe people just care about price and not certifications. The company is also now trying to do safety testing for AI systems. It just rolled out a new standard called UL 3115, a, quote, structured framework to evaluate AI-based products before and during deployment. That kind of standard requires a lot of companies and regulators to buy in and for there to be a way to even reliably safety-test AI at all. And then there's the structure of UL, which, well, we'll see. It's complicated. But structure aside, we'll get there. I had to ask Jennifer if she gets to watch stuff explode in the testing labs, because to me, that seems like the best part of working for an organization that sets safety standards. Turns out a lot of stuff blows up in those labs, and you'll hear Jennifer say her office often rattles because of it. There are also other kinds of explosions. Right at the tail end of the Biden administration, UL got tapped to be the lead administrator for a new cyber safety program. It was supposed to set a standard for connected devices, the Internet of Things. But as the Trump administration came to power, power-mad FCC Chairman Brendan Carr came up with reasons, which, of course, he never actually articulated to anyone, why any company related to China is now somehow a threat. That apparently includes UL, which has safety labs in China because that's where the electronics are made. So the UL lost its position as the lead administrator of the CyberTrust program. I asked Jennifer about that directly, since that's pretty much a microcosm of how our government is dealing with safety, tech, and China right now. There's a lot going on in this episode. I love it when we get to bring hidden systems to light here on Decoder, and I think you'll like it. Okay, UL Solutions CEO Jennifer Scanlon. Here we go. Jennifer Scanlon, you're the president and CEO of UL Solutions. Welcome to Decoder. Thank you, Nilay. It's such a pleasure to be here. I'm excited to talk to you. Some of my favorite episodes are when we demystify a thing that everyone takes for granted, and the UL logo is one of those things. Absolutely. The UL mark is on billions of products, and yet everywhere I go, people look at me and say, what exactly does UL do? My understanding is that you just drop things off of cliffs and see if they explode. Is that day-to-day? We do have people who drop things off of cliffs and see if they explode. But really, every single day, we have 15,000 employees around the world working for a safer world. And they are testing, inspecting, and certifying products. And they are also creating software to help our customers manage their risk and compliance environments. You run a big testing facility. Describe some of the tests that are done and who gets to do them and what's some of the wildest tests that you do are. Yeah, I always like to say we break things, we blow them up, we light them on fire. And so if you were to walk into our testing facility here in Northbrook, Illinois, in Europe, in China, in India, anywhere in the world, what you'll see is, first of all, large electrical panels that are there charging, discharging products, batteries, and seeing what fails. Watching a lithium-ion battery the size of my thumb blow up is pretty terrifying. It's amazing how wide that blast will go. So we do a lot of inherently unsafe things to test product safety. I think my most favorite test, the team, I wasn't there, but I got to see pictures of it. We stacked 2 million, 2 million soda pop cans in our large-scale fire testing warehouse and then dropped a lighted piece of paper in the middle to see what would happen. And to this day, I don't know if we were testing the aluminum, the labels, the contents, but I do know the test failed. They were supposed to cave in and kind of collapse upon themselves, and they instead exploded, and it took a number of days to clean up the 2 million failed soda pop cans. And that's what we do. We protect our customers. They needed to know that what they thought was going to happen didn't happen. Oh, no. What's the most dangerous test that you've gotten to be there in person for? The most dangerous tests, you know, I do think are hazardous location testing, which is in Northbrook. Everyone's door in my office is kind of right above it, and you'll feel a little shake. And you really think, how bad could it be that aluminum in a combustible dust environment sparks? Well, if you think about that, you know, you're out on an oil rig. You're out in, you know, some factory. You know, a lot of lives could be lost. And so, you know, while the test itself is well-controlled, it really makes you think about, you know, the lives that are at stake of what we're doing every single day. Do you ever bail out of boring meetings and just go blow stuff up for fun? I don't think the engineers would let me, but they do enjoy it when I come visit because I do ask a lot of questions that I'm always fascinated by the things that we do. I think they would let you. I got to be honest with you. I know a few engineers. I think they might be like, yeah, we'll set something up for you. Oh, it's great fun. 15,000 employees. That's a lot. The company started a long time ago as underwriters laboratories, right? It was fire insurance companies that needed to make sure electrical devices weren't going to burn down your houses and they could write fire insurance. Is that still the basis of the company? How does that work? Yeah. You know, I like to say that the basis of our company was to address the safety of the technology of the day. And at the time, 1894 World's Fair, right? You know, on the edge of University of Chicago, where both you and I have a bit of history. We, the Underwriters Electrical Bureau did bring our founder to Chicago to help do some primary scientific research on the safety of electricity to write standards about how that electricity should be used, both manufactured and embedded into products and installed and safely used in buildings. And then perform public advocacy, educating people on the new technologies. So fast forward, certainly with the electrification of everything, the energy transition, AI data centers, you know, electricity and electrical safety continues to be a primary worry and a driving force. But there's lots of other new technologies of the day that we continue to help keep our customers safe. What are the other technologies that you're mostly focused on? You know, some of the most current ones, AI safety and the ways in which AI is being embedded in products and the ways in which humans engage with the safety of AI and products. And so our newest, as we call it, an outline of investigation, which is a precursor to writing standards, was published in November. And it's all around the safe use of AI embedded in products. I want to come to that. That feels like a very meaty subject of conversation here. There's a lot of AI safety debate in our country and in the world. So I want to come back to that. Absolutely. I just want to start with some foundational questions that I have. One of them is where the authority to tell the industry what to do comes from anymore. Right. So when you had a bunch of insurance companies saying we won't pay your insurance claims if the thing that burned down your house wasn't UL certified. Well, that provided an awful lot of incentive for people to go get that testing done to pay for it. At the time, UL was a nonprofit. A lot of that has changed since then. Right. Where is the authority or the incentive to participate in the UL process come from anymore? It's a really important question. And relevance is a really important strategic concept that we focus on a lot. You know, who does it matter to if your product has been certified to a UL standard or even another standard? We certify to over 4000 standards. Only 1500 of those are actual UL standards. So there's others, authorities having jurisdiction and standards development organizations globally. The importance of this is that governments, certainly insurance companies, underwriters, even today and in the U.S. tort system, that becomes very important. They are looking to ensure that either, you know, what they're underwriting is safe, what various agencies of governments around the world deem safe. And how do you Industry, particularly the tech industry, which doesn't like to listen to anyone. Oh, they don't. I started my career there, yeah. How does that conversation go? Yeah, you know, it goes like this, and I'm going to give a really great example. Let's talk about e-bikes and in particular, you know, e-mobility devices, but e-bikes in New York City. You know, five years ago, let's say, you know, I believe there were a couple dozen people who were killed in New York City, and why? You know, overcharging of the lithium-ion batteries. Lithium-ion batteries have a different chemical composition. The thermal runaway happens faster. The chemicals are more difficult to put out. And, you know, I hate to say in a typical house fire, you have a couple minutes to get out. With a lithium-ion battery fire, you have fewer than 30 seconds to come out alive. So you've got this problem. People are dying. You've got this problem, which is people are excited to use e-bikes because they're an affordable mode of transportation. They are a very useful item. So how do you balance this? We at UL Solutions heard from a number of customers, worked with our not-for-profit partner, who is our largest shareholder, UL Standards and Engagement, to write three standards around the safe charging, the use of batteries, and just the ways in which lithium-ion batteries were installed in e-bikes. Three standards. We went to New York City, worked with the mayor's team there, as well as the fire services team there, to ensure that those standards were written into New York law. Now, once a standard is written into local legislation, what that means is if you're a bicycle manufacturer, you're not going to manufacture a different bike or a different charger to sell into New York City than you would in Chicago or Toronto or LA. So it starts to proliferate. The good news is, is since those standards were adopted in New York City, deaths have dropped by 75%. So there is a real need, you know, for the safety of humanity in these standards, and then that becomes picked up by other authorities having jurisdiction, other, you know, communities like those other cities I named, or even, you know, like local private campuses, universities have expressed interest in, you know, what are these standards? How do we think about ensuring that a dorm doesn't catch on fire? Or so that's kind of the genuine approach, the authentic approach to how this happens. There has to be, you know, the safety science that shows what, you know, what the answer could be and should be. And then there has to be, you know, a recognition that that need is real and that it helps promote that trust between, you know, those authorities having jurisdiction, those governmental bodies, and the citizens and the users of products, you know, within their jurisdiction. It's interesting because the choke point there is retail, right? The city is not going to let you sell a bike without the certification because it's deemed the bike without the certifications to be dangerous. Is that consistently the kind of incentive that makes people adopt the standards of the certifications, that you have to stop that there's enforcement somewhere? Not always. And, you know, I'll go back to, and we're going to talk more about this AI standard, UL3115, but that started with our customers coming to us. And we see this a lot. Our customers saying, hey, you know, as a manufacturer, and I spent almost 20 years in manufacturing, if there is a standard that we should adopt and that we know our competitors will adopt that levels the playing field and creates a marketplace, you know, and creates a consistent marketplace. So frequently our customers come to us and say, we see this happening. Help us think about how this new innovation, this new technology should be considering what the safety science is. And that becomes the precursor to writing a standard. And frequently our customers don't even wait for the standard to be written. They start using that outline of investigation, as we call it, to guide their product design and innovation so that they're more confident coming back to that insurance question, you know, of if something happens that they will not have a, you know, a failure in safety. I want to come back to the notion of customers because UL has been restructured and you as CEO of UL Solutions have taken that part of the company public. That's for profit. I'm very curious about that set of incentives and what that means. But it just strikes me every time I talk to somebody who runs a standards organization, and we talk a lot in the industry, whether it's Bluetooth or HDMI, there's just some element of being a politician that's involved in that. Right. You would not think of Bluetooth as a deeply political organization, but they have a lot of unwieldy stakeholders who are pulling in different directions. You're describing it as we need to create a market. We're not actually, you know, with HDMI, maybe you want a feature that no one else wants, and that's a political problem for that standard. It doesn't seem like you have that same set of pressures. How much politicking do you do? We really don't do politicking. What we do is in the standards development process, it's a consensus process. So as I said, our customers frequently come to us with the need for a standard. AI data centers are a great example. You know, moving to 800 volt DC is a very significant energy need and safety challenge. So how do we start building standards around that and then kick that over to UL Standards and Engagement, who's actually the standards development organization where they convene technical panels and follow a consensus-based process. So there's some pretty rigorous approaches to that standard development and the consensus grounded in science base. Now, getting that standard adopted by governments, you know, does take, and again, our Standards Development Organization does this, the not-for-profit. They are involved in ensuring that the right attention is given to the opportunity to adopt those standards and spend their time promoting why it's, you know, why it's an important need, why it's a good idea. Let me ask about the structure then, because you are describing the inner relationship between the three parts of UL. UL started off as obviously one organization. How is UL structured now? You've got the three divisions. Why the change and what are the divisions here? Yeah, we were not-for-profit from our founding in 1894 until 2012. And while we had, you know, as I described earlier, that we were founded to do the safety science research, the standards development and the public advocacy immediately following the World's Fair, companies started coming to Underwriters Laboratories, asking for their products to be tested, inspected, certified. We did that as a not-for-profit, but charged for that until 2012. In 2012, our trustees, the trustees of the not-for-profits realized that our competitors, many European, who were founded with, you know, kind of similar histories as not-for-profits, had the opportunity to both do a better job funding the not-for-profit side, but also unleashing, you know, in an increasingly competitive environment, really officially being for-profit and kind of unleashing that, you know, that type of energy. So in 2012, we split the two. I joined in 2019 as CEO of the for-profit with the relationships back to the not-for-profit around the standards development and the research. So today they are structured as three separate entities. The standards development organization is the shareholder of UL Solutions. So when we went public in 2024, two years ago last weekend, they were, it was a secondary offering and they received the full set of proceeds to fund their endowment for their standards development and for their research institutes. So we've got a separate board of trustees, a for-public board of directors, four trustees sit on our board of directors. So there is, you know, good strategic relationship. And I think that that's very important, but we are run completely separately. We have to take a quick break. We'll be right back. Welcome back. I'm talking with UL Solutions CEO, Jennifer Scanlon about UL's complicated structure. When you say completely separately, I'm actually curious to tease out the lines there. So there's the three organizations. There's UL Standards and Engagement. There's the UL Research Institutes, and there's UL Solutions, which you're the CEO of. Solutions is a public company, but you've got the trustees of the nonprofit on your board. How much do they get to tell you what to do? So, you know, I was the company’s CEO prior to joining UL Solutions, and I don't see any difference between this board and my previous board because there's a distinction. You know, my previous board, I had Berkshire Hathaway as my largest shareholder, and they did not sit on our board. And I was well-trained that as CEO, as the management team, and as the board, we serve all shareholders, not a single shareholder. And I treat that in the same vein here. There is all of our shareholders deserve, you know, equal attention and, you know, kind of duty of care, duty of loyalty to all of them. And there is strategic value in having the right strategic relationship with the not-for-profit, and that value goes to all shareholders. So that's the way we think about it. That's the way we treat our board meetings. That's the way we treat our management decisions. I'm very curious about the sort of commercial incentives you have around the for-profit part of the organization. And I understand you had a lot of competitors that became for-profit testing labs, and I know the decoder audience enough to say, well, that obviously corrupted them, right? They're just selling marks now and selling more marks makes them more money, and maybe the testing standards have gone down. And I'm just curious how you balance that. Like, I hear that from our audience a lot, right, that the financialization of everything has corrupted everything, and the trust is gone because everyone's just chasing dollars. You run a public company, you've got shareholders to deal with, you're talking about them. How do you manage that? Like, you probably could lessen the standards and sell more certifications, and that would probably be better for your shareholders in the short term. But obviously, there's the long-term of the brand and what it means to people and the nonprofit to protect. How are you balancing all of it? You know, we've been around for 132 years, and we still speak the words of our founder, which are, know by test, state the facts. And if we were to ever deviate from, you know, the highest quality standards, if we were ever to deviate from the highest quality science, it would erode the trust that our customers have in us that we've built for 132 years, and our business is trust. And so I fervently believe that we have to continue, you know, kind of this long-term view of, I use the phrase, growth and relevance. Grow as far as, you know, our influence and our ability to advise our customers and support them, but remain relevant. And the only way you remain relevant is if you maintain that trust. When you say customers, you don't mean consumers, right? You don't mean the end user. You mean big companies, governments. How do those customers express their preferences to you in the market? Yeah, our customers, you know, we have three segments to our business. So industrial, which customers tend to be selling their product in the B2B space. Consumer, where customers tend to be selling their products in the B2C space. And then our risk and compliance software segment, where those tend to be our largest, you know, multinational, global, and strategic accounts. So our teams are out there working with the new product development teams, the quality and compliance teams in all of our customers. And our customers express their needs as they're going through their innovation cycles. We frequently have a line of sight into their product roadmaps, a line of sight into how they intend to use technology differently in innovation. And I say frequently, innovation without safety is failure. And I think our customers feel that same weight. They don't want to fail. They don't want to have a product launch that's going to harm somebody, either in that industrial environment or that consumer environment. And so it's a really open, honest dialogue because we're there to help them. And sometimes helping them is giving them news that they don't want to hear. But it's incumbent upon us to tell them, you know, this is, these are the facts. This is what happened in the test. And now you have to go back and do something about it. And we can't advise them on how to redesign their product. That would be a breach of that trust. We have to stay agnostic and test when that product sample comes in. So I'm very curious about that, right? You said customers don't want to make products that hurt people. The tech industry says that to us a lot, right? And in particular in AI, they say this to us. They talk about alignment and safety all the time. And then we can all see the reporting about what chatbots are doing to consumers. Where is that balance? Is it all just industrial applications? We don't want the AI to run the elevators wrong. Or are you looking all the way to what are the model capabilities? You know, for us, we focus on products. We focus on product safety. And, you know, functional safety of products would be, you know, when you embed software, let's say, you know, in an electric vehicle, you don't want to turn the radio on and have the brakes slam, you know, because the latest software download, you know, changed the if-then-else statements and you find yourself in a safety problem. So similarly with AI, you want to make sure that AI is not creating, you know, functional safety challenges, but you also want to ensure, and what we're hearing from our customers, is that they can portray, profess that there's trust in the models. So our UL 3115 came from customers coming to us and saying, you know, and it's a great example, you know, a child's toy. How do you know that the data that was used to train that AI that's embedded in a child's toy, that that data was fair, that it remains private, that it's transparent, that there's, you know, lack of bias in the algorithm. And because all of that determines how that product actually performs. And so that's the perspective that we have. But back to your first comment about, you know, the technology industry being very resistant to others, you know, setting standards or, you know, guidelines or regulations, we fervently believe that third-party independent testing, inspection, certification leads to better outcomes for society. I mean, I can point you right now to AI-powered children's toys that are completely off the rails. And I would just bring that back to what is the enforcement mechanism? What is the choke point? There's no New York city that's going to say you can't sell teddy bears in our town unless the AI has a certification. I don't think that exists in some of these markets. Are you, where are you finding the enforcement or the incentive structure that makes them participate? You know, it's early days, and I completely agree with you that, that we've got to get our arms around this. There are a number of standards development organizations around the world, you know, not just UL, but, you know, IEC, ISO, others that are coming together and saying that, that this is, this is necessary. This is important. We will continue to advocate various governments, authorities having jurisdiction, tech industry associations, others to continue to pursue this. But it is, you know, indeed, you know, a, I'm going to call it an uphill battle where, you know, the tech industry likes to have their own, you know, their own approach and, and will cloak themselves in intellectual property and proprietary standards. And, and I get that. I started my career at IBM. I understand, you know, the value of tech and IP, but I'm a lifelong safety freak. And I really believe that, you know, some of this stuff could turn products into being inherently unsafe and we need to do our best to prevent that from happening. Let me ask you the other decoder question I ask everybody. And I want to dive into it using that framework. How do you make decisions? What's your, what's your framework for making decisions? So my personal framework is ground in data. I am a data person. And, you know, I think you need to have enough data and pressure test it to make a good decision. I believe organizationally in empowering people that, you know, if your job is to run X, then you should be grounding in data and making decisions around that. And then there's a certain level that they should, you know, potentially get bubbled up to me. But, you know, a lot of times, you know, I, I think the people closest to, closest to the customer, closest to the decisions need to make that. There is one set of decisions here at UL that I will never overrule, and that is the scientific decisions that our scientists, our engineers, our lab technicians make. Every once in a while, a customer is not happy with a report or a decision that, that we have made and it can get raised to me. And I think my team has the confidence to know that I will never overrule a scientific or engineering decision. Yeah, that seems very important, right? That's the heart of the enterprise. It's to protect the sanctity of the testing. I'm, you know, in the, in the context of AI, but even in the context of batteries, which I want to talk about at length, actually, it feels like the market is getting farther and farther away from wanting to comply, right? So I'll give you the example here. You know, the Biden administration really pushed for AI safety and they had a set of standards that they wanted to promulgate. President Obama was on the show talking about the need for AI safety. And his comparison was, very explicitly, we failed to regulate social media for people. We're not going to screw that up with AI. And we at least want the labs to publish their testing at the very least. Trump administration showed up. All that is basically gone, right? That Biden era EO is no longer in effect You know, I agree with you. There's not any, you know, kind of top-down forcing function right now to bring them to the table. Are you engaged with OpenAI or Anthropic or XAI or Meta? We're not directly engaged with OpenAI or Anthropic. We certainly have done a fair amount of work with Meta through the years and most of, you know, the hyperscalers and, you know, more on the product side. But these continue to be topics of conversation that our chief scientist and our PhD researchers in AI are out there promoting and continuing to try to, you know, push the rock up the hill. Yeah. You mentioned your new standard, UL 3115. It kind of, it's a pretty wide-ranging standard, right? It's everything from data centers to consumer applications. I think the first two products certified under it are out or the certifications were received. They're like building control applications, from what I understand. That to me is, okay, we're going to certify a building control application to make sure it doesn't go haywire and like turn up the heat in all the units or whatever a building control application might be able to do. All the elevators are going to go crazy. This is just a philosophical question. These AI systems are fundamentally non-deterministic, right? They're not predictable in the way that they operate. And that actually is what makes them powerful, right? There's the bad side of hallucinations and them posting to their own weird internal Facebook that they built for themselves. And then there's the good side of, oh, that means they're creative. Like they can write, they can do software development in a way that a deterministic system really could not do before. How do you test that? What is the mechanism of testing whether an AI-powered building control software is always going to do what it says if the engine powering it is inherently unpredictable? AI models really rest upon that predictive modeling. But our focus is not on, you know, getting into the black box of the code. Our focus is on establishing, you know, I'm going to call it over 200 criteria around how internally when they're making decisions about their code development, how they should think about bias, how they should think about transparency, how they should think about fairness and privacy, and, you know, the whole range. When you say think, is it the models thinking or is it the people making the models thinking? Well, I think it has to go into how the people designing those models, you know, so how are they building out, you know, what is the veracity of the data source that they use to train the models? And so that's outlined in our standard of how should they make decisions about that. And so I love that you focus on how decisions are made. When I look at UL 3115, I think that it is a standard to help guide those decisions as AI is being embedded in products. The big opportunity AI right now is software development, right? The cost of producing new software is dropping precipitously and may drop to zero because the tools are so good at it. And tools like Claude Co-Work and OpenClock can just go do things for you all the time, which is really fascinating. That means the number of providers of software, the number of providers of AI-empowered software is just going to skyrocket. Like it might get away from you in a way that when you describe the market-making capability or the market-making function of UL, everyone is going to get this certification, so we're all on the same level playing field. Well, if the playing field is vast and it's a bunch of teenagers writing applications in the race things who don't care about you, it might just totally get away from you. How do you think about that balance of there's going to be big players who want to participate and get the logo mark, and then maybe there's going to be an entrepreneur saying, you know, I can make you this building control software much cheaper, who never actually comes to you. Yeah, and I think that's where getting back to our customers and what they're looking for, where that comes in and how they're going to level the playing field of their competition. And at some point, you know, the end consumer does speak. You know, I was in manufacturing for 20 years. I don't want unsafe, you know, kilns or metal presses happening, you know, AI-powered in my environment. And so there's a point at which you're going to want that verification, that validation, that endorsement that what you're installing in your industrial environment or what you're bringing into your home as a consumer is safe. And so that's where I do think the end user at some point has a voice because they're going to decide, do I want to buy this product or not? And we have, you know, plenty of tests that we do that have absolutely nothing to do with an actual regulation, but have to do with the fact that our customer has decided that this is important for their brand, for their end consumers, and that that drives the demand for what we have to offer. The other dynamic that's happening in AI specifically is that the models themselves are getting ever more capable and the idea that you need to build a specific AI application that's a wrapper around the model that forces it to do what you want. Who knows how that's going to play out, but you can see, well, maybe actually I just need a subscription to Claude and I don't need a subscription to some application that is powered by Claude because Claude can just do it for me. If those companies aren't engaged with you, how does this work? If Anthropic and OpenAI and the rest aren't engaged with you, how does this work? Yeah, it's a really great question because it comes back to, you know, one of the concerns or questions that I actually have about AI, it comes back to that veracity of the training data. And the more, you know, back in my coding days, garbage in, garbage out, the more garbage that gets in to train these models, the more difficulty you have in trusting whether those models actually have the efficacy into the future, you know, or do they just spiral upon themselves and become useless. And so I believe that that should really be appealing, you know, to these development companies around, you know, does that model have the longevity to continue to provide, you know, the answers, the intelligence, the information that is grounded in something that is actually, you know, true and correct. Let me ask you straight out, do you think they care? I hope they care because it's, you know, it should be self-preservation for them to care. I mean, they seem to be doing pretty well without caring. That's why I'm asking. Yeah, well, there's short-term and long-term, but, you know, we'll see how this plays out. You mentioned that the pressure to reign in, be more safe, have more control for AI might come from other governments, other organizations. Maybe it's the state. Where do you see the most pressure on making AI safe come from right now? You know, it's interesting, and I know you've spoken with, you know, some of our large customers recently, and I think it's coming from those large multinational global customers who, you know, care deeply about how their products are used in environments and, you know, want their relevance and longevity to be out there. So they don't want to find themselves in a situation of failure. When you talk about those customers, are they coming to you and saying something like UL 3115, this is what we need it to say so that when I test it meets our needs? Is that how that standard is developed? No, it's, they've come to us and said, we need a standard. Help us think about it. And so as we start to develop it, we bring them into a room and then we've got our, you know, our PhD AI researchers in there with them. And it's a, you know, it's a dialogue grounded in science with then a consensus of, okay, we think that this actually will really help us. Let's make sure that that's in there. PhD AI researchers are very expensive lately. They are. Can you pay at the top of the market for those folks? You know, we've built a small and mighty team in this and we feel very good about their thought leadership and what they've contributed. Yeah, I'm just, I'm curious because that's the other arms race. Like the, I look at this from the outside and I say, no one can keep up with these labs. They're paying all the money that even a competition between them doesn't seem to be keeping them in check. The idea that they're all going to sign up for a literal checkmark from UL that says they're safe when they're all racing to an IPO. I'm just very curious where that pressure is going to come from. I don't, I don't know if it's going to come from an industrial manufacturing supplier at this point in time. I think it might have to come from a government. Yeah. And I think we've got to keep pushing this rock up the hill. It's still early days and it's important to figure it out. We're going to take another short break. We'll be back in just a minute. Support for Decoder comes from Adobe. For every big idea, your documents folder tells a story. Let's say you've just finished pulling together a brief. So you hit export on final version .pdf. But then you open the file and you immediately notice a typo. Several versions later, you're exporting final v4 dot actual final draft dot pdf. Adobe Acrobat can save you the digital clutter with PDF spaces. It takes your documents and turns them into a living project that you can engage with, get insights from, and collaborate with others on. You can gather all your files into one workspace and have a whole conversation with your AI assistant about it and ask questions to get deep insights about your project. You can even invite people to your PDF space and let them The other piece of this, as you mentioned, the standard covers data centers. There's a lot of tension, political and otherwise, around data center buildout in this country, everywhere else. There's just the electrical component of it, right? If you're going to do a lot of electricity in a room, you probably want a bunch of UL-certified components in there. Is there more than that in UL 3115 as it relates to AI data centers? Yeah. Now, UL 3115 is just really around AI embedded in products. With the AI data centers, there's 70 other standards that we test to today around the safety of the electricity, the components, the chillers, the DC current coming in, the inverters, all of that. And then there's a whole host that we're hearing from our customers that with the rapid change in the amount of power, the rapid change just in the thermal dynamics of GPUs versus CPUs, the rapid change in the way that you're going to put a megawatt of power into a rack or that you're shifting the water cooling. So there's a whole set of new standards way outside of UL 3115. We've had two summits with customers on AI data center summits on how they're thinking about their needs for standards and data centers and how we can rapidly help them continue to develop on their innovation pace in ways that they can feel comfortable will be safe in the future. Do you think they're going to slow down their buildout goals in order for these certifications to take hold? No, I think they're expecting everybody else to pick up the pace. Fair enough. Let me ask you what the other race condition, because again, I think it would be great if everything was certified and everything was safe. And I kind of look at the markets that we're in, and there's just an explosion of things all the time. And the one that really strikes me is everything with a battery in it. I go on Amazon, we're profiling one of these companies here at The Verge all the time. If you've got a lithium-ion battery and a high-efficiency motor and a dream, you can start a company that makes 500 products today. We profiled some of them, Hodo and Fantacart too, that have just sprung up and they make tools. And the other day I saw one of those companies had a like a Sawzall, like a lithium-ion handheld Sawzall, which is just a lot of power. If you're gonna put that much torque in a little motor, that's a lot of power you're gonna draw. And I look at these companies, and they're all, they're obviously all based in China, and whether or not they have UL certification is irrelevant to the consumers buying all these products, because products are cool. They're legitimately cool products. And there's, in a race of innovation happening there. And it's all just on Amazon. And Amazon doesn't seem to be enforcing any of these standards at all. How do you think about that? How do you think about the, just the prevalence of high-powered lithium-ion batteries everywhere without the consumer demand for your certification? Yeah. You know, first of all, Amazon is a great customer of ours and you can drop down and see if something's been UL certified. So I, I do want to highlight that. They should make that more prominent. I think you should probably tell them to make that more prominent. They are a great customer of ours. And indeed, innovation is fast. Batteries are exciting and dangerous. And we continue to work with, you know, customs agents, various authorities having jurisdiction, and our customers to help educate, you know, in particular, if you're importing into the U.S. markets, how to keep those lithium-ion batteries safe. You know, a great example, this was about a decade ago when hoverboards were exploding. Boy, did we cover the exploding hoverboards at The Verge. Yes. So the Consumer Product Safety Commission came to UL Standards Engagement, UL Solutions at the time, and said, can you very rapidly write a standard and help us get our arms around this? And we did that. And again, it helped with the safety. And then one of the key areas that we have is market surveillance and anti-counterfeiting. So we are constantly working with customs agents and, you know, competitors who are putting the UL mark properly on their product, who will highlight a product that's in the market that's not meeting the codes and the standards. And we've won some significant lawsuits around these cases where there are unsafe batteries, uncertified situations, you know, when they're not in compliance with the law. Yeah. Amazon and UL together are suing some e-bike manufacturers or selling on that platform with fake UL certifications. Exactly. You have to catch them. So you have a, you have an enforcement team that's actually scanning Amazon for fake UL certifications. We have a team that works with, you know, anyone who wants to highlight that they think that there's a unsafe situation or a counterfeit UL mark, our team responds. Can you scale up fast enough to meet the flood of new products that, again, a lithium-ion battery, a high-efficiency motor and a dream, you can start a company and make 500 products tomorrow. Can you scale up to meet that flood in terms of testing? You know, what we feel, I mean, we can absolutely scale. We've scaled all over the world. And we'd like to say, you know, we meet our customers where they are. So if you're doing innovation in China, we've got our testing labs in China ready to go. If you're doing innovation here in the United States, we've got our labs here ready to go. If you're manufacturing anywhere in the world, our field inspection team will visit your plant four times a year to ensure that you're manufacturing in accordance with the standards that we test it to. So we've been growing. We'll continue to grow. Do you make the case to Chinese manufacturers, hey, if you have this UL certification, you will make more sales. Absolutely. Absolutely. And there's data showing that the U.S. consumers actually care about this. Yes. And manufacturers in China all across Asia, they know if they want to get their product into the U.S. market, they need to follow the safety standards, and we're there to certify for them. I know you just made an acquisition to expand your testing presence in the EU. How big is your presence in China compared to the United States and the EU? Yeah. Our reported revenue last year, and we report revenue by point of customer. So if you're a U.S. customer, but we're testing your product in China because you happen to have an innovation center there, we will report that in the U.S. So last year, 42%, I believe, of our revenue was point of customer in the United States. 25% of our point of customer is China. About 17% is EMEA, and the rest is kind of the rest of Asia and the rest of the world. So China has been very important for us. We've been in there 40 years. We've got a joint venture partnership, and we have independently wholly owned labs as well. And we work very closely with a significant number of Chinese manufacturers to help them get their products, you know, to markets all over the world. The relationship in China has been the point of contention recently with the Trump administration. During the Biden administration, the SEC launched an eCoffin Cybertrust Mark, which was supposed to certify IoT devices specifically as being safe. UL was supposed to lead investigator, sort of write the standard for that. Brendan Carr, who is well known to listeners of this show and my other show, The Vergecast, is the current chair of the FCC. He has a lot of ideas, Mr. Carr. And he decided that your relationship and your work in China somehow was corrupting. Something happened, which I'm dying to know what exactly happened. UL is no longer participating there. And then Trump donor, his company is now the lead investigator. What was that conversation with the Brendan Carr FCC like around the Cybertrust Mark? You know, when the Cybertrust Mark was first announced, I mean, we're proud. We have been, you know, we're a proud American company. We've been here for 132 years. You know, if our government asks us to serve, we, of course, will step up and serve and support whatever they need. And so we were really pleased with the work that we were able to do to help set up the lead administrator, you know, as the lead administrator, kind of help set up the parameters of that and work closely with the FCC. When the FCC decided that they wanted additional requirements from the lead administrator, we realized that we weren't the best fit for that. And we, you know, very kind of easily transferred that intellectual property and that work back to the FCC and they continued down their path. What were the additional requirements? So those requirements were really around how they wanted to run the program in the future. And it was, you know, just a set of requirements that we didn't feel that we were the best participants to do that. That sounds very bureaucratic and administrative. I'm looking at Brendan Carr. He basically accused you of being beholden to the Chinese government. Did you ever respond to that directly? How would you respond to that now? You know, the way we would respond to it is we have been very transparent about our operations, our relationships all over the world, and we continue to be so. Yeah. Brendan is not a subtle man. He doesn't do things in the shadows. He says you're beholden to the Chinese government. You're saying that is absolutely not true. And it was enough for you to walk away and say we don't want to be a part of this. I think that where we all landed is the right answer for all of us. Similar to the FCC right now is banning a bunch of Wi-Fi routers simply because they're made overseas. You've obviously looked into this with the CyberTrust It says consumers prefer UL products, and I hope that's true. But then we at The Verge cover, I don't know, cameras for your house that have just gaping security holes in them, right? Where there's just like live feeds streaming to the whole internet at large because there isn't a security apparatus or an updates apparatus. We do see that with routers. We've seen a lot of hacks with consumer-grade routers. I'm just wondering where that extends to, particularly in software, right? You buy a power strip, you can see the logo on the back of it. Or maybe Amazon will at least show you the logo, and maybe you'll still buy the cheaper one because you don't really know what it's for. With these software products or, you know, these hardware products that are running a lot of software, it's not right. So how do you make that case? Yeah, you know, and we do have a service that focuses back to that functional safety of that embedded software or that efficacy of that product being connected to the internet and the cybersecurity of that. And so there are standards around that, and there are ways to approach it. But I think what you're highlighting is an opportunity to make consumers more aware of what they should be looking for and demanding as they purchase their products. Do you think that this is just a market problem? I think maybe this is what I'm coming back to over and over again throughout this conversation, is I really wish the consumer market demanded more of these companies. But, you know, that's just a collective action problem. Like, I think it's totally rational for most people to just pick the cheapest power strip that Amazon has on the first page. And I can't really blame them for it. At the same time, maybe we don't have a federal regulator who's going to step in and say, okay, to keep everybody safe, we're going to demand this certification. Maybe we don't have insurance companies who are going to go demand them of Amazon. And then when it comes to software, it seems like the tech industry in particular is utterly resistant to anyone telling them what they can do. And the idea that, you know, a UL certification for firmware updates on your camera is on a cadence is just maybe the hardest sell of all. So if it's not going to be the consumers that do it, and we have a government that seems checked out of it, this is what I keep circling. It's what I was most excited to talk to you about. It's what I keep circling back to, is where is the pressure come from for people to participate in a safety program? Yeah, and I think this is, to me, one of the exciting pieces of when we went public and funded the endowment for our not-for-profit. And so we've talked a lot about the standards development organization, UL standards and engagement. We haven't talked so much about the UL research institutes and the areas where they're focused. And one of their institutes is focused, you know, really on AI safety and how should, you know, the world be better educated of, you know, what would be considered safe and where they should dig deeper. And so I think there's a lot more to come on not just the research around that, but also around, you know, what is that step to raise the consumer's consciousness of the fact that, you know, as I like to say, by the way, if something's free, you're the product. So if you're giving your data away free, you know, back to social media, you know, you're using it and it's free. You're the product. And so how do you protect consumers, you know, from that? And I think it's a really important concept and I still think it's early days on this and AI. You're in the business of selling safety. I think it's a fair way to describe what UL does. Do you think that the way that Dario Amadei or Sam Altman talk about AI alignment and safety is effective? Because their pitches, if you don't let us do whatever we want, we might kill everyone in the world. Yeah, I think, you know, I think they're trying to ground, you know, in science and engineering and certainly, you know, in different ways to use AI and, you know, different models and, you know, and LLMs or, you know, one approach, but there's lots of others. I think that it's probably a false choice to say, you know, let us do what we want and therefore we'll prevent this from destroying. I think you need both. When you say both, you mean outside testing and validation or government regulation? What do you mean by both? All of the above. I think that, you know, the tech industry, it would be ideal if they came together and said, here's what we believe collectively will help, you know, keep the world safe and then we'll adhere to that versus letting each one just go off and, you know, follow whatever path they think is best. If you had to if you had to pick a structure for that to happen in again, you manage a complicated safety structure. So I'm just asking rhetorically. If you had to pick a structure for that to happen in, does that look like a government regulation? Does it look like an industry body? Does it look like a nonprofit that controls a for-profit testing center? How would you how would you design this? Yeah, and I think where it's heading is, it's more of the standards development organizations and the industry bodies coming together. And because they will be the most knowledgeable about what should work, you know, I mean, you, you always want that deep industry expertise when you're developing any type of safety standard that then moves into regulation. I think if you start with regulation tops down, you don't always get to the right answer and it's not always grounded in, you know, the science and the engineering that it needs to be. So I would advocate, you know, industry groups with standards development organizations. Jennifer, what's next for UL? What, what should people be looking out for? You know, for us, it's going to be this continuation, as I like to say, of growth and relevance. Um, we will continue to be on the forefront of innovation and continue to find ways to make safety relevant for whatever innovation comes next. You know, I can, uh, geek out and get excited about quantum for a second, um, as something that's, you know, the future, you know, extension of what's post AI or, you know, what does, you know, makes AI better. Um, but those are areas that we continue, uh, to try to stay involved in and, um, you know, think about not just, you know, the electrical safety of 132 years ago, the electrical safety needed in data centers today, but what is coming next? I do like that even though it's been a long time since you've been at IBM, you brought it back to quantum. That's a very, that's very IBM of you. I really appreciate that. It is. I have to say, I'll give a little IBM shout out, um, because I love Arvind, but I was walking through O'Hare and they have their IBM quantum chandelier sitting right there next to the dinosaur. And I mean, I skidded to a halt while I was pulling my luggage. I'm like, oh my gosh, it's a quantum chandelier. And it's really exciting because, you know, I am here in Chicago and we at UL have been involved in, um, you know, creating the quantum ecosystem that, that Chicago has, uh, been promoting and, and we're excited about what's next in that set of technologies. So we can talk about that another time. The second someone ships a working quantum computer that does economically relevant tasks, we'll have you back to talk about it. I don't know. I don't know when that's going to be. Closer than we think, I hope. That's a bold prediction. Thank you so much for being on Decoder. This is great. Thanks. I'd like to thank Jennifer for taking the time to join Decoder and thank you for listening. I hope you enjoyed it. If you'd like to let us know what you thought about this episode or anything else at all, drop us a line. You can email us at decoderattheverge.com. We really do read all the emails. You can also hit me up directly on threads or blue sky. We're also on YouTube. You can watch full episodes at DecoderPod. We have the same handle on TikTok and Instagram. They're a lot of fun. If you like Decoder, please share it with your friends and subscribe wherever you get your podcasts. If you really like the show, hit us with that five-star review. Decoder is a production of The Verge and part of the Vox Media Podcast Network. Our producers are Kate Cox and Nick Stat. This episode was edited by Kabir Chopra. Our editorial director is Kevin McShane. The Decoder music is by Breakmaster Cylinder. We'll see you next time. Support for Decoder comes from Adobe. Life is unpredictable, and that means you need your projects to adapt with whatever gets thrown at you. That means mastering the ability to pivot and collaborate with others to reach your goals. Adobe gets that, which is why they made a tool that's just as flexible as you are. PDFSpaces and Acrobat. Your PDF files are no longer static. Instead, they're living documents that flex with you and your project's needs. Learn more at adobe.com slash do that with Acrobat.