The New Legal Stack: How AI Is Rewiring In-House Legal | Ron Klain (Airbnb) & Ross McNairn

Hi. Thank you.

Sorry, I'm late. Couldn't find it.

Yeah. Why legal and why now? Well, let me start by saying before I get to that a little bit about Airbnb. It's a very familiar brand, but not everyone knows exactly what we do. We're a global technology company that operates a platform that allows 5 million hosts to accommodate 2 million guests every night, 365 nights a year. And as Genevieve mentioned, around the world, virtually every corner of the globe, every part of the world, except where we're prohibited by law or other policy considerations. The platform, as our competitors, is relatively new, but the form of hospitality that we provide or that our hosts provide is ancient. All three of the Abrahamic faiths, Judaism, Christianity, Islam, have teachings that people should welcome guests into their homes, and virtually every ancient culture had a mandate that guests should be welcomed into people's homes. The Greeks, most notably, had something called xenia, the idea that Zeus would punish people who didn't welcome guests into their homes. And when the film The Odyssey comes out this summer, you will hear a lot of talk about xenia in that film and whether or not Odysseus and his men were welcomed or Telemachus later on. So, the idea of welcoming guests into your home is an ancient idea. The platforms, of course, make it easier than ever. And that's what we do at Airbnb. It's complicated. So, I have about 240 lawyers who work for me around the world, half of them in the United States, half of them scattered between Europe, Asia, and Latin America. And so, why legal, why now? Well, for us at Airbnb, I would say that AI tools provide a critical aspect of running a global legal department. Even though we haven't had them that long, I could not imagine doing my job without them at this point in time. In terms of the way they facilitate our legal research, our contract life cycle management, our legal workflows, they're absolutely indispensable. But what I'd say is to answer the question that's on a lot of people's minds is that legal AI at this point in time and for the foreseeable future is not going to replace lawyers who do creative, thoughtful, nuanced, sophisticated legal work. It replaces a lot of the work that lawyers are doing that isn't really legal work, that's more administrative work, or that's kind of repetitive work that you could train a high school student to do. Like imagine a task that lawyers are often given. Here's a pile of contracts with commercial agreements with our partners and our vendors. Go through here and make sure each agreement has an indemnity clause and make sure it looks kind of like this. Well, lawyers do that task, but it's not really a legal task. It's a pretty simple repetitive task, and contract lifecycle management tools can do that task much better than any human can. But if I say to a lawyer, a more experienced lawyer, 'We're going to go negotiate an agreement with FIFA to be one of their sponsors. You know what our business objectives are in this agreement. Try to get a mix of commercial, policy, and PR concessions that are best balanced to serve our interests.' There's no AI tool that could do that now. And so I think that lawyers who are experienced and good at their jobs should not find AI threatening. They should find it as a useful adjunct. But it is going to be a challenge for the lower end of legal work. No question about it.

Yeah, obviously, AI intersects with our legal department in two ways. One is how we use AI to practice law, and the second is how we counsel our company on its use of AI to serve our guests and hosts. And so, we're using AI increasingly to power our customer service in 11 different languages around the world, and to improve our guest experience when they're searching for the right place to stay. Because people want different things when they search for a place to stay. Some people want certain amenities. Some people want to be near the Eiffel Tower. Some people want to be far from the Eiffel Tower. So, using AI to improve our search results is part of it. And we as lawyers have to counsel the company on the issues raised by that. And that includes GDPR compliance, its impact on our DSA status, and all sorts of other legal restrictions around the world. And so, we are both AI-powered lawyers and AI lawyers every day. And you know, picking, I agree with Ross, picking the tools carefully is important. Not becoming dependent on one foundational model is also important as prices rise and fall, and legal departments are cost centers in most companies, and so I'm on the hook both to expand our use of AI as a tool, but not blow the budget. And so, having some flexibility there is good. And we're very mindful, particularly in the US, because the rules are a little different than the US, about privilege. And what our lawyers are putting into these models, and how it's protecting our attorney-client privilege. And so, I think people have to be thoughtful about all this. And it is a powerful tool and a highly useful tool, but it has to be used carefully. And it has to be done in a way that's compliant with the existing web of laws that regulate how we handle customers' information.

Well, I think that right now we're kind of doing what we have to do to do what we have to do, and just try and keep an eye on what that's costing us in terms of token consumption because I have members of my team coding with Claude tools for us to use in legal. And we've built a product review tool that allows our colleagues in engineering and design to upload Figmas of new products and then it manages the workflow of how that clears past our privacy lawyers, our payments lawyers, our consumer protection lawyers, our regional and local lawyers, and so there's some token consumption building that. What it's enabling us to do right now is provide legal advice more quickly, not by cutting out the legal reviews, but by getting rid of the dead days when a privacy lawyer finishes a review and sends an email to a payments lawyer and says, 'You're up now.' And the payments lawyer doesn't look at their email box for a day. So, we're trying to get rid of some of the dead time in these product reviews and speed up our product reviews, enable our company to speed up our product releases, and that has its own benefits, of course. But I do think in time managing these costs is going to be an important issue for in-house legal departments, but we're trying to do more work more quickly without adding heads. We're not cutting heads, but we're not adding heads. And we're trying to do more work more quickly with the same people, and so things that improve efficiency and the speed of our work and the quality of our work are very useful tools.

Well, I think it's a great question. I'll be brief because we're out of time, but I think the construction of new data centers has become one of the most controversial issues in the United States. We're about to face our midterm elections in almost every part of our country and almost every race, people's views on more data centers is a central issue. I think we need to make sure that we're doing this in a way that's environmentally responsible with regard to power consumption and water consumption and the environmental impacts on the local environment of building a large data center. And this is going to be an issue of political contention for many years. But I do think there's a way to do it responsibly. Responsibly on the power side and on the community impact side, but it has to be built into the equation. It has to be a subject of governmental action and also industry action.