AI Has an Insurance Problem

[MUSIC PLAYING] US-Mexico trade talks hit a roadblock in Washington, plus Tesla's new roadster takes flight, literally. And we'll look at AI's next big obstacle, insurance. The worry is that the sheer growth of the industry means the claims are potentially outpacing the coverage limits, particularly for the labs developing this technology. It's Friday, October 16th. I'm Luke Vargas for the Wall Street Journal. And here is the AM edition of What's News, the top headlines and business stories moving our world today. [MUSIC PLAYING] Commerce Secretary Howard Lutnik has thrown a wrench in US-Mexico trade talks, stalling progress toward a potential tariff deal. We're exclusively reporting that Lutnik won't agree to lower tariffs on Mexican steel, aluminum, and automobiles, which all fall squarely under his directive as Commerce Secretary. Journal Trade Reporter Gavin Bade has more. So for weeks, the US and Mexico had been zeroing and on a deal to reduce some of these levies. US tariffs on its steel and aluminum currently sit at 50% while US tariffs on Mexican automobiles are at 25% in particular. Lutnik and some of his allies in the domestic aluminum sector in the US have objected to lowering tariffs on what are called derivative products, finished goods that are made from steel or aluminum. And they are responsible for the vast majority of the jobs in the US aluminum sector. And so they've kind of drawn a hard line at not reducing those tariffs under 50%. Obviously the Mexican government not very happy with that outcome. On cars, Mexico is hoping to lower a 25% tariff to the 15% levie enjoyed by Europe and some East Asian countries. But Gavin says that's also facing resistance from Lutnik. And now what we're hearing from sources is that prospects for an agreement are not very good before the midterm elections here or maybe even before the end of the year. That throws the US-Mexico trade relationship even into further doubt. This narrow tariff deal on steel, aluminum, and autos, it was supposed to pave the way for larger discussions on renewing the US-Mexico Canada agreement or at least renewing it for US and Mexico. All of those things are now in doubt. So we'll just have to watch this space and see if they can get this over the line. In the months to come. The Justice Department is suing California to block the state from counting mail-in ballots from some 20,000 voters that the Trump administration claims may not be US citizens. Filed less than three weeks before the midterms, lawsuit seeks an injunction against counting those ballots until citizenship status can be verified. California Secretary of State Shirley Weber pushed back, arguing that the federal government provided no basis for its suspicions and calling the legal action in an emergency of the Justice Department's own creation as statewide voting gets underway. Meanwhile, voters' economic concerns are putting intense pressure on Republicans across an expanding battleground map. The nonpartisan Cook Political Report predicts that Democrats could flip nine to 19 house seats more than enough to erase the GOP's current five-seat majority. However, local dynamics present a more complex picture as seen in New Hampshire, where Republican governor Kelly Aod is holding onto a steady lead in the typically blue state. Across the border in Maine last night, Republican Senator Susan Collins and Democrat Troy Jackson held their final debate in one of the nation's most closely watched Senate races, the two clashed over skyrocketing utility bills with Jackson citing high fuel costs linked to the war in Iran, while Collins fall to the state's Democratic leadership. So, you know, Senator Collins, keep on talking about how our electric rates were risen, because she doesn't want to talk about how our oil rates have risen. Thank you, Senator Collins. Every state is affected by the war in Iran's impact on oil. Not every state has the high electric rates that Maine has. The main race is one of seven Senate battles that the Cook Political Report rates as a toss-up. Six of the seven are currently Republican held. Well, as promised, Fed Chairman Kevin Warsh is holding firm in his refusal to signal the central bank's next rate move. Here he was speaking to the head of the IMF at its annual meeting in Bangkok, where their microphones might benefit from some repositioning. I don't prejudge meetings. I don't believe in saying now what I might do with the meetings that we have in the balance in this year. But that was the situation. It's a pointed half of the audience or more. They can all turn it off. It's a new chapter at the Fed, and we're not in that business anymore. Most of Warsh's colleagues have penciled in at least one more rate increase this year. The Fed's October meeting falls less than a week before the midterm elections with any decision to raise rates or hold them steady, potentially drawing more political scrutiny than normal. Warsh also predicted that the AI build out would ultimately boost the economy's production capacity without necessarily pushing up prices, saying the US was on course for a quote, positive supply shock. Well, speaking of shocks of the visual sort, Tesla last night unveiled its new Roadster sports car using on-board rockets to help a test model climb a curved ramp and invert itself mid-air before making a controlled landing. Flying cars are real. Yeah. The car later used those rockets to jump over five other Tesla vehicles and what the company called a kickflip. Patrick George is the journal's Detroit Bureau chief. The car did what Musk said it would do. He's been promising a car that could fly in the hover for years now using SpaceX rockets. And on Thursday night, it did exactly that. It also pulled a zero to 60 time of less than one second. That exacts claimed it'll do north of 230 miles per hour. Patrick said that Roadster's unveiling will come as a relief to enthusiasts, some of whom put deposits on the car nine years ago, though they'll have to wait a little longer for delivery. On stage in Texas, Elon Musk admitted that it goes into production in 2028, though he added, actually, for real. That's probably a bummer for anyone who put down a positive up to $250,000 back in 2017. But that's often how things go with Tesla. Then there's the question of how this fits into Tesla's business model. Musk has said the company is pivoting to autonomous cars and AI and robotics, so the Roadster will be the last human driven car that Tesla ever makes. In theory, if it's expensive enough, it'll be a profit machine for Tesla, the way that a lot of high-end sports cars from Lamborghini or Ferrari are. Musk didn't respond to shouted questions at the event about the Roadster's price tag. In battling rising fuel and other costs, trucking giant JB Hunt is turning to price increases. Speaking to investors, executive vice president Darren Field defended that move, saying that the company had already done its best to trim expenses and the customers understand the inflationary environment that it's operating in. They're not surprised by our discussions around our costs. I think as this year went on, early on in the year, there was some concern and a lot of pushback from customers on price increases, but as the year has gone on, those conversations have become more realistic. JB Hunt's CFO forecast that higher operating costs will extend into next year. Coming up, we'll look at how a string of AI safety incidents is exposing a gaping hole in the industry, the lack of insurance. That story and more after the break. The insurance industry has seen a thing or two. That's how it can price in the risk of a car crash, a hurricane, or even an act of terrorism. But as the journal's Jean Eagolson reports, the industry is struggling to keep pace with AI as the potential damage caused by emergent threats run into the tens or even hundreds of billions of dollars. Jean, you know the insurance industry, backwards and forwards, is the issue here in protecting you and me from AI-related risks or in providing liability for the labs who are unleashing AI into the world or as your story centered a little around both. It's really both, so there's liabilities here that are new for the labs. And there's also liabilities for all companies using it and for us as employees or as customers of those companies. So we've just seen this dizzying array of incidents recently from rogue agents hacking into websites to allegedly chatbots encouraging folks into suicide to AI hallucinations by law firms. So AI has this incredible potential to help us and help the economy but it's also got a potential to do harm. And that's what the insurance is worried about. I mean, just to be clear, these sorts of AI risks that you're just describing are things that aren't covered under as a general liability or cyber insurance that I'm sure many companies have. So it's not clear at the moment exactly where the liability falls. It's not clear yet tested how much of that is covered by existing policies and the insurers are considering excluding some of these risks as well. And even where they are covered, the worry is that the sheer growth of the industry means the claims are potentially outpacing the coverage limit. It's particularly for the labs developing this technology. Especially when we're talking about civilizational risks in some cases. I'm guessing that's probably not the majority of what we're talking about here but some of the damage here can be. really widespread. - The damage can be very widespread because you've only got a few labs so you have pretty concentrated risks and also just thinking about the type of harms here. If say you had a hack that took a company down, a big bank down for a long period of time, the economic damage from that could be huge. So that's why insurers are looking at this very seriously. They're trying to quantify the risk. And as you say, nobody is willing to underwrite what they call the tail risk. So civilization ending, nobody is going to be sure a policy for that. - I guess there's no one to get paid out or to pay in that kind of situation. Though presumably most of what we're talking about here is a niche that some insurer is going to be willing to step into now. - Yeah, definitely. There are startup insurers now who are just focusing on this. So they're writing policies specifically trying to target this new AI risk. The big insurers are also looking at this. So this is an area that's changing very fast. It's a market insurance market that's growing very fast. But as we stand now, they say there are big gaps. Particularly when you look at the sort of open AI and thropic those ones, their liabilities are way ahead of their insurance coverage. - Gene, I'm just curious. We've got some large AI lab IPOs coming up the next few weeks, months, years I'm sure as well. Could that play a role in sort of forcing those biggest companies to figure out a durable insurance solution that works for them because surely their investors are going to want to know that they've got protection against these risks we've been talking about. - Yeah, definitely these IPOs are increasing scrutiny of the sort of financial safety nets. These big company tens. So if you look at something like an anthropic or open AI, they have incredibly deep pockets. So to some extent, they can just cover these liabilities themselves, but definitely regulators, lawmakers, others are looking at insurance, are questioning the gaps in insurance because insurance could provide kind of reassurance that an outside company has looked at these risks and said, "Okay, I'm willing to underwrite them." And another wrinkle with these companies is that previous generation of tech companies, your Facebooks, for example, they had this legal protection called Section 230 that gave them quite a lot of immunity over being sued over the content on their websites. Now that doesn't necessarily apply to this new generation of AI companies according to Moody's analytics, because the companies themselves are contributing to the content. So they may not have that shield on their liability. A lot of this is still open to testing and courts, a lot of it's undecided. - And we may see, it sounds like from your reporting, we are already seeing a push by some of these tech companies to try to secure themselves some sort of US government liability backstop. I'm not sure if that's like equivalent to replicating Section 230 for them or something a little bit different, but we could look to Washington here, possibly for something to shake up in the next few years. - Yeah, I think it's going to be pressure on governments around the world to address this issue. So you already have, for example, government backstops for risks like terrorism, which was just too big for the insurance industry to absorb after 9/11. Oh, in the UK, you've got a government scheme for flood risks. So I think a lot of people in the insurance industry, as well as on the tech side, are looking at these systemic risks from AI and saying, how could we structure something that would be able to cope with that? - Reporter Jean Eaglesham recovers the insurance industry for us at the journal, Jean. A pleasure, thanks so much for bringing us this story. - My pleasure, thanks so much. - And finally, with the weekend, right around the corner, I get it. The last thing you probably want to be thinking about are the dreaded Sunday scaries. - Miriam Webster defines it as, quote, "A feeling of anxiety or sadness one may feel when the weekend is almost over. Either because the work/school week is about to begin, there is no more time for fun or relaxing weekend activities or both." - That's workplace reporter Ray Smith, who found that his workloads mount in job security ebbs, the scaries are getting worse. - More than half of U.S. workers say they experience a Sunday scaries. That's according to an October survey of 2,100 people by Flex Jobs, a job platform. The most commonly cited calls of the Sunday scaries for those people who said they experienced them, having a list of unfinished tasks hanging over them. - To avoid that pre-monday stress spiral, Ray said that one solution, maybe to try to get more done today. - When I talk to people about combating this feeling of dread, some say they really just try to at least get their thoughts out on paper or on note on their phone like early Sunday. They also try to set limits or guard rails on how much time they're gonna spend on work. One woman I talk to told me she purchased an hourglass and she uses it to stop her from working any longer than an hour on Sundays. And so far, it's kind of working for her. A psychologist recommended maybe consider spending a little more time at work on Friday. So you can be done with the week and not have to like carry it over into the weekend or an alternative to that is getting to work a little earlier on Monday. If you feel like you wanna get a head start, or if you wanna catch up, rather than using your Sunday to do that. - And that's it for what's news for this Friday morning, it's not Sunday yet, don't worry. Today's show was produced by Hadi Moyer and Daniel Bach, our supervising producer is Sandra Killhoff, and I'm Luke Vargas for the Wall Street Journal. We will be back tonight with a news show. Otherwise, take a load off, have a great weekend. Thanks for listening. (upbeat music)