Generating Ai Value webinar

[00:00:00] Jonathan Bench: Griffen looks like we can get started. We’ll give everyone a minute or two to get in, I think, but we’ll let everybody know who’s here so far that we are, we’re on our way, and we’ll get started in about a minute or two.

Okay, Griffen, maybe we’ll get started and we’ll do, we’ll do our intros. I think my favorite part of this Of this webinar is seeing everybody’s AI chatbots is showing up. So we probably have 50 percent live people, 50 percent chatbots that are going to be picking this up and parsing it later for interesting tidbits.

Or maybe it’s even people who can’t attend and they’re sending their, uh, sending their chatbot instead to listen and, and then summarize it for them later. So maybe we’ll start with introductions. Griffen, do you want to kick it off? Let everybody know who you are and why you’re interested in this topic.

[00:00:49] Griffen Thorne: Sure. Uh, good morning. Good afternoon, everyone. My name is Griffen Thorne. I’m a partner in our Los Angeles office. Um, I’ve been practicing law since 2015 and I started a big firm doing intellectual property litigation. Uh, back in 2015, and then, uh, started picking up privacy law, uh, in, like, 2016, probably, and at that point in time, there, it was basically a brand new area of the law, uh, we didn’t have, like, internationally, just the, the privacy law regime was sort of coming into place, and so I got to practice in the early days of that and through today.

Um, today, most of what I do is corporate transactions, just a variety of different types and a variety of different industries, but I do a lot of intellectual property licensing and privacy law type work still, and, um, this stuff is becoming more and more. I mean, as everyone who’s watching knows, this is becoming a hotter and hotter topic.

So we’re seeing projects, uh, yeah. Very, you know, new and unique projects, like on a very frequent basis. And so we thought it would be a good idea to host this webinar and, you know, get everyone up to speed on the legal side of things.

[00:02:03] Jonathan Bench: And Griffen’s good chance, I think, for you to talk for one second about the, the project you and I talked about where you, you’ve had, I’ve used AI in, uh, in a transaction recently, and I’ll talk about that in a minute, but maybe you could talk about your, uh, your foray into the space as well.

[00:02:19] Griffen Thorne: Well, I mean, I’ve certainly I’m a little more skeptical of things like as they currently stand. I’ve seen a I drafted contracts and we’ll talk about this a little bit later today. And I’ve seen a lot of issues with them. Not to say it can’t be done the right way, but it might be. Difficult right now, uh, the technology I think will get there, but it’s I don’t think it’s quite there yet.

Um, at least without like some hand holding Um, but I I did a project Uh, I want to say a couple of months ago where a client was developing an ai tool To use on its website sort of like a chat bot And I put together the policy and the terms and conditions for it. Um, and it was integrated. I can’t I can’t remember the underlying AI tool that was used off the top of my head, but you know, it’s it’s complicated because you’re wrapping in privacy policies in terms and conditions of both the client and then the underlying system.

So which can often be. Much different from, you know, just general client privacy policy issues. So, yeah, we, this is all something that we’ll probably talk about later today as well.

[00:03:23] Jonathan Bench: Thanks. And on my end, uh, Jonathan bench, I am a partner in our Salt Lake City office and Griffen and I spend a lot of time working on clients and including high tech clients, uh, domestic us and international transactions.

So a lot of interesting things, you know, you take normal, uh, Business transactions, and you push them across international borders, which is exactly what technology can do so easily. Uh, and, and then all of a sudden you’re in, you’re in 20 jurisdictions, and you’re dealing with customers, and, uh, you know, terms and conditions, all government policies in, in all the nations where your software has been deployed.

And so it, it Turns, uh, what is a difficult project in the U. S. into a, an even more difficult project once you’re going international. Um, so I’m a, uh, I guess I, I haven’t considered myself a tech lawyer, Griffen, but the more I hang out with the tech folks, the more I start to talk about myself as a, as a bit of a tech lawyer.

I’m definitely a techie lawyer. Uh, I love technology and grown up with it, uh, like most of us. Uh, I feel at this point, as we’ve been preparing this webinar and as I’ve been thinking about, uh, other AI presentations I’m going to be doing or overseeing this year, uh, particularly with the Young International Lawyer Group that I’m a part of, uh, we have a conference in Mexico City coming up in May, and I’m, I’m overseeing a panel on, uh, on AI, two uses of AI, one as a, as an asset for business and the other as a, Kind of a transactional amplifier.

Uh, and so I’ve been thinking about this for quite a while. Uh, I’d say two months ago as part of a transaction, I started using, uh, I think I used Clod for this just because I hadn’t used Clod. I used ChatGPT quite a bit. And so I, I used Clod and I anonymized a 30 page contract. Uh, and then I dumped it into Claude and said prepare, prepare a closing agenda for me for this transaction.

And when I call it closing agenda, Claude didn’t call it closing agenda, and so it was, it was more like a checklist. But what I wanted was the, all of the, all of the ins and outs, all of the outstanding deliverables, you know, you and I are familiar with this. When we get into big transactions and there are, uh, you know, you’ve got How many conditions precedent before you close you’ve got 20 different things that need to happen by closing Then you’ve got things post closing and and it turns into a big big headache to keep track of all of that And so I had to use the guy used it probably took me 15 minutes So it took me a while to anonymize the contract But then once I got I dumped it into Claude I think about 15 minutes and maybe five different Questions later, five different prompts later, I actually got what I consider a pretty good closing agenda that would have taken, you know, me or an associate easily, easily an hour probably to go through and pull out all the important things, get the formatting right.

And it was, it was, it was workable. So that was very interesting and certainly got me excited about it. Um, yeah. So let’s talk about, uh, you know, the technology in general, because a lot of people, I’m sure people who have read up on it, there are a lot of YouTube videos, a lot of different viewpoints, you don’t get lawyers thinking about new technology that much, unless we’re forced to, and, uh, and so I hope that today we can give a little bit of our viewpoint, uh, from, from the legal side, certainly we, we’re, we are lawyers, we think like lawyers, and we see, we see everything as lawyers, and, and I’ll try to give a little bit of the business flair, and I’m sure you will too, Griffen, for the, the kinds of, uh, Uh, opportunities that that will certainly see, uh, come out.

So I guess in just in terms of introduction to this technology, uh, AI is, uh, it’s really about parsing data. It’s about labeling things. And so, um, is the reason why. It’s taken this long in the in the history of the internet where we have all this information available. We have Data scrapers that can go through and pull out terms and then It is amazing And I don’t I’m not a I’m not a computer science major and I haven’t I’ve only watched a lot of really interesting videos on this on how the the technology actually works, but you need very Massive data sets.

And then I understand that in order to really deploy these models, well, you need a ton of computing power. If you have just a minimal amount of computing power and a lot of data, you’re not going to get the returns, especially on the generative AI that needs to keep producing prompts and producing prompts and kind of leading out from the questions that are being asked.

So it’s it, uh, this inflection point in history, you know why things have happened now at this point in time is in large part because We have, uh, we have a lot of computing power. We have GPUs that can be produced and linked together, and we’ve seen this in blockchain. We see this in supercomputers, and certainly we see it, uh, the applications now for, uh, for generative AI models.

Um, in terms of importance, I thought this was interesting. I’ve been paying attention a lot the last few months to the news as, as we’ve been preparing for this. LinkedIn came out with a top 25 fastest growing jobs in the U. S. in 2024 and the thing that caught my eye is number eight and number 10 on this top 25 list of fastest growing jobs in the U.

1. are related to A. I. One is an A. I. consultant, which is more of someone who really understands how A. I. works and how it can be applied. And then number 10 is an A. I. engineer. And so the fact that, uh, A. I. You know, we’re only, what, 13, 14 months after, after the official launch of, of ChatGPT and now we have, uh, the top two of the top 10 jobs in 2024 are, are dealing with AI and that’s just in the U.
2. I don’t know about other, other economies, but certainly is interesting. Um, I mean, you’re in California, Griffen, I got to imagine that the, uh, everyone, everyone breathes technology, uh, all the time there. So, um. I don’t know if you’ve, if you’ve heard anything about, about that, or if, uh, or if we’re still early enough that this is a, you know, a forward looking prophecy.

[00:09:12] Griffen Thorne: Well, I mean, I’m in LA, so it’s not Silicon Valley, but I mean, the big Thing that was recent with respect to AI was the Hollywood strikes and the fear that AI would replace writing and possibly even acting jobs. Um, and I think like to some extent, uh, that kind of thing is inevitable in every industry. Um, it’ll, it will definitely change how we do work in the next five, 10, for the rest of our career, basically.

And I, I I think it’s difficult to come up with a job that’s not eventually going to be affected by this. So, yeah, it doesn’t surprise me that even though some of this stuff is relatively new, it’s already creeping into the top, you know, 25 positions. You know, it’s just, it’s an explosive technology. And I think that the difference between 2019 and 2024 is, and then the difference between 2024 and 2029 is going to be Massively different, right?

So I think, you know, we, nobody really knows what’s going to happen, but we’re seeing just an explosion of this stuff that everywhere and in every industry. And so there’s just going to be massive

[00:10:15] Jonathan Bench: shakeups. Just this week there was an article in the Economist talking about how AI has a potential to transform emerging economies again.

Right. So, really interesting to get the international flavor on that. Um, I’m gonna reference a, uh, a, a very well known ai, um, I guess he’s one of, one of the, one of the, not founders of ai, but really one of the, uh, one of the people pushing this forward. His name is Andrew Un and he is a. Computer scientist, um, now at Stanford, uh, associate professor there, but he’s really a, uh, you know, he’s a technologist and he’s, he has a venture capital fund, I think at 200 and 250, maybe, maybe it’s 200 million VC fund that’s just focused on AI.

And, uh, he’s been, uh, uh, he’s got some great videos on YouTube and I, I highly recommend him if you are a person who’s trying to figure out what the AI applications can be for virtually any industry. Uh, he’s got some great videos. One of them, a couple that I watched, uh, you know, these are under 20 minutes or so, including one that he did as a presentation at, I think it was Berkeley and, uh, just fantastic, really solid understanding and, uh, and really kind of, uh, his, his, one of his catchphrases that is, has caught.

Uh, you know, caught on now is the A. I. Is the new electricity. And so he uses that to explain that, uh, you know, electricity has many uses. You know, everyone needs electricity now, and it makes businesses better, right? It’s something that foundational. And so he sees A. I. As, uh, as doing that as as being that transformational as electricity has been, you know, through the industrial revolution.

Uh, so I want to run through a recent timeline. I thought it was very interesting. Pitchbook, uh, Pitchbook put out a timeline of some of the most important events in 2023, uh, in the AI landscape. So, and, and these are telling because I, I feel like, uh, you know, news moves so fast and, and we, you and I read a lot, try to keep up with what’s going on, you know, U.

1. and international business. There’s just so much. Um, and so it was. It’s actually fun to review this because I lived through all of these things. I can remember when I read all of these headlines, but seeing it all together on a timeline was, was very interesting. So November 30th, 22 was when chat QBT was launched January 10th, just a month and a half later, Microsoft took a 49 percent stake in open AI, a 10 billion investment.

Then a month later, Google invested this February 3rd, Google invested 300 million into anthropic, which runs clawed AI. Uh, and Google also launched its own. chatbot barred. Then in March, Salesforce announced a 250 million dollar fund to invest in ethical AI. So only three months after Chad GPT, we have a lot of concern around the ethics of AI.

And I think you mentioned Griffen with the, the strikes in, in Hollywood. That was, uh, I remember reading about that now that you mentioned it, that it was a big issue is how, uh, how secure are people going to be and how quickly will this technology potentially replace so many, um, white collar jobs? Really?

You know, I mean, that’s what I think about it is this is, this is kind of a, Uh, it’s an intellectual crisis for a lot of people who are in the knowledge economy and, uh, and people who are in the more blue collar areas are, are less worried and less likely to be affected, at least in a way that would jeopardize their jobs.

We can talk about that more, but I thought it was very interesting, uh, way to think about all of a sudden this crisis among, uh, among, uh, white collar workers. So that was, that brought us to March, April. Amazon launched, uh, what’s called the Bedrock Tools for building Gen AI apps. May 16th, uh, Sam Altman and AI experts get pulled in before Congress and, and GBT 4 was launched.

Then in May of last year, NVIDIA’s stock tripled in value, uh, in eight months, largely from the demand in the AI industry. It became, for a little while, it had a trillion, no, it was a three trillion dollar market cap, I believe. So it went from a one trillion dollar valuation to three in a period of eight months, just on the back, mostly on the back of AI chip demand.

Uh, and then in July Metta released Lama two, which is an open source AI model. And, and then September 25th, Amazon invest 4 billion into anthropic. So a whirlwind, uh, I don’t know if you have any comments on that Griffen, but I mean, it was, uh, I felt like I couldn’t keep up with it fast enough, but it’s really fun to have it laid out like that and see how quickly those tech majors really jumped into, uh, you know, some significant investment dollars.

In the technology. Yeah, I mean,

[00:14:39] Griffen Thorne: I don’t know. It’s where there’s money to be made there. People are going to flock to it. And I think that’s also the technology just is what it is. You know, there you kind of breezed over it, but there was the push to, um, essentially put a pause on developing this stuff to figure out how to do it the right way without causing danger.

And I just, I don’t, I don’t, well, first of all, that didn’t happen. Right. And second of all, that’s that, like, I just don’t think that genie can be put. Back in the bottle, you know, on a temporary basis because it and I mean, it’s kind of like a cliche at this point, but if it’s not developed here, it’s going to be developed somewhere and we can put all the regulatory burden on the industry that we want.

But there’s countries that aren’t going to do that. And in fact, we’ll fuel the fire, so to speak. Like, you can be sure that China and other places are going to push on this stuff as fast as they can. Um, I don’t want to use the term like arms race, but it could be that like that in the future. So, um, this stuff is moving so fast that we really don’t even know what’s going to happen next.

And I mean, obviously, that was the fear with with some of these people who are asking for some sort of regulation. But, um, you know, working in a lot of regulated industries myself, uh, representing clients in those industries. I’m skeptical that it can be done the right way and just seeing like privacy and how that’s regulated.

And in fact, like very over regulated in a lot of places. Um, it’s it’s a little concerning to think what that could look like in the future. And there’s there’s just a lot of concerns, right? With talking about ethics. Um, if you’re. Well, we could probably come back to that in a little bit, but, um, yeah, why don’t why don’t you get back to the outline and we can move into the next topic.

[00:16:21] Jonathan Bench: Yeah, sure. So, I think the next natural question as a, as a business lawyer, where I’m sitting is, is how are people going to make money with this? Uh, you know, clearly the, the easy answer is, well, if, if you have. A few hundred million dollars you can, uh, you can put in the work and develop your own, uh, uh, you know, large language model.

You can, you can hire the techs and you can, you can go through that process. That’s already happened and, and there’s, and that’s one reason why, um, I’ll talk about it a little bit. It’s one reason why, uh, the U. S. government, uh, Especially is very interested in what’s happening at the at the big tech companies and AI and interested in making sure they don’t strangle out the smaller, um, you know, smaller companies.

So I’m going to share a see if I can get this in the right way. I’m going to share two slides and I happily admit that these are not my slides. So I’ve kept Andrew Ng’s kept his name here at the bottom, right? So, The so there are two things that I want to mention is this is thinking about the business case for this so he talks about And the way that AI is broken down and you see that The the little orange dot so he is how he describes this.

He says the the darker color Uh, is where we are right now, and then the lighter color is where, is where the technology is going to expand in the next three years. It’s not a very long timeline, um, and so you can see the growth. But he says, this, this idea of supervised learning, of labeling things, that’s the, that’s the foundation for, uh, for generative AI.

And supervised learning has been happening for For decades, and that is how we get targeted ads for us when we’re online. This is all the things that say, well, if, if someone is interested in a, then they will be interested in B or they’re likely to be interested in B. And so you can imagine in terms of, of the way, especially in the advertising, you know, online, that is, that is really, as you think about why we get so mad with the way our data is used online, that is, that is what’s happening.

It’s this supervised learning, the labeling of things and, and how. People can build up customer bases or take advantage of people’s interests. Generative AI is still comparatively small in terms of what’s going to be happening in the next few years, but still, uh, he’s predict he’s predicting massive growth.

So, I this was one of my favorite charts that that he put together. And so if you look on on the far left, um, that that is where we’re talking about ads in web search. That is that is where the tech has really been deployed so far. And and he made a really good point in one of his videos that Tech is, um, tech is, is just the tip of the iceberg, and tech is really only a small part of the overall economy, and the genius of, uh, of AI, and especially generative AI, is going to be in applying it to regular things, things that we would never, or that we wouldn’t necessarily associate with, uh, with needing generative AI, or, uh, or AI at all.

So I’ll, I’ll use two examples that, uh, that he used. One was this idea of the pizza, the food inspection. So he talks about a Uh, a company that’s producing pizzas and a frozen company pizza, big, you know, massive operations. He said it’s about a five million dollar problem they’re trying to solve. And the big issue was, how do they make sure that the cheese is distributed evenly across all the pizzas, so that the customers are happy.

So this was a big deal for them. And, uh, and so he said, you can use, uh, you can use a I to train, uh, to train your, you can train the model on photos. And so he said there, you know, there just aren’t a lot of pictures of pizzas from overhead, right? You can imagine trying to trying to get enough data together where, um, Where you can start training the AI model to say, okay, this is good, or there’s or there’s cheese missing on this section or it’s moved.

I mean, it’s I worked in a pizzeria as a as a teenager. And so for me, this is kind of a really funny but interesting application for it. And and, you know, big commercial applications for it as well. So any I guess one of the takeaways for me is that any Uh, any industry has data, right? And every bit of data is important.

And, and, uh, and the data, those who collect the data in any industry are, and track it and understand what data matters are the ones that are going to be able to use AI in order to Improve their processes whatever that means for them. It could be could be any number of ways and so for me It’s just it’s just really interesting.

You know taking the idea that if AI is the new electricity Then every company can use it. And so then the question is what data is the company producing? Even if it’s even if it’s low tech, right, even if it’s like he’s got a pure material grading, you know, how do you train? You know, how do you train the AI model to detect fabric defects or scratches on screens or scratches on phone on phone protectors, you know, something like that.

So any industry has practical application for how this can be be utilized. And I really like that. The other example he used was a shipping company, massive global shipping company, and they came to his team and said, We want to optimize our fuel consumption, you know, because in the, in the, uh, in the oceans, the currents are always moving.

There are winds, lots of different factors. And so, um, there’s a, there’s a really cool app, Griffen. I don’t know if you’ve seen this one. It’s, uh, I can’t remember the name of it, but it tracks all of the maritime traffic. It might be just called marine traffic. I think mass, an app that shows every ship in the world that’s on the sea.

Uh, based on its transponder, if it’s agreeing to, uh, you know, to transmit its data. And so you can imagine with, with, uh, I don’t even know how many thousands or tens of thousands of ships are on the seas on any given day. If you can take all of that data and parse it for how quickly they’re moving when they’re checking in, then Uh, you know, then ships can use that data to navigate their, the optimal course.

Uh, and he said that they were able to capture, I think, 10 percent fuel savings for these big ships as they were, as they were moving across the ocean. So I, uh, you know, I geek out a bit on this stuff, you know, on kind of what, what are the practical commercial applications for this? And when I hear things like, it’s good, it’s good for every industry, you know, same kind of thing where people ask me about blockchain.

They said, is blockchain, is this useful? How, you know, how can it. Is it, is it, does it go beyond crypto coins and tokens? Like, what is the, what is the actual practical use? And as you dig in more and you understand, there are, you know, where technology is, uh, we’re talking about distributed networks, we’re talking about, uh, really capturing inefficiencies in, in businesses.

And for me, that’s like, that’s, uh, I love thinking about that stuff. It’s so interesting to me.

So I think that’s all I’ve got on, um, kind of on the technology, on the future of it, where I, where, where we see it’s going to go, going to go from here on, um, I think, uh, you were going to talk about privacy and confidentiality, which I’m, I’m really interested in, in hearing about. Sure,

[00:23:18] Griffen Thorne: so I want to, we can shift now into some more of the legal analysis, as opposed to sort of the tech and the business side of things, because we’re lawyers, and I think that’s what you’re all logged into today.

And I want to talk specifically about privacy, confidentiality, and then people who deal in regulated professions like ours, right? Or work in regulated industries, and maybe they themselves don’t have a license and some of the impacts of all this stuff, but to do that, I kind of need to talk about what are the main laws we need to be concerned with.

And on the privacy side, there’s many of them. I mean, You know, if you’re in the banking industry, there’s, there’s laws about consumer data and if you’re in the, there’s just various different things, but in general, the big ones I like to think about are the European Union’s General Data Protection Regulation, GDPR, I’m sure you’ve heard of it, and then in the United States, like, the biggest one essentially is the California Consumer Privacy Rights, I’m sorry, California Consumer Privacy Act.

There was a Consumer Privacy Rights Act that kind of modified it, but it’s like they’re similar acronyms. But yeah, that’s one CCPA. In the U. S., we don’t have one unified federal privacy law. There have been attempts to do it, but it just never happened. Probably because it would create a massive new agency and, you know, you’d, you’d For generally don’t see people on the right wanting to agree to things like that.

And so what ends up happening is a state like California passes a law that is so robust that it effectively applies businesses all across the United States and even globally, sort of like GDPR does. And so, um, it’s almost as if. A federal privacy experts out there might not agree with me, but it’s almost as if one is not necessary, right?

Another sort of series of laws that just generally apply are laws governing data breaches. So every state in the US has a data breach response law. When I was starting in this field, there was like 40 states with them, and a bunch didn’t have them. And they’re all sort of the same, right? They all kind of look similar.

They have similar features, but they can depart in drastic ways. Uh, for example, like. First of all, we talked about personal information, right? So if a business has some personal information of mine, they have certain obligations in the event there’s a breach, which is generally defined as somebody, you know, having unauthorized access or, um, uh, taking of data, right?

So, Every state defines that personal information category differently. Some it’s very narrow, like social security information, whereas others it’s very broad. Biometric information may be included. California, I think, includes license plate imaging. So there’s a lot of different categories, and this depends on the state, which is where lawyers have to, like, contend with a lot of potentially different.

Factors here. And so if there’s a breach, um, generally, these laws require notice the consumer and then some states might require things like providing them with, uh, uh, credit monitoring services at the company’s cost. And so, as you can imagine, this can be incredibly expensive. I worked on probably 25 different data breaches at my last firm, and, you know, I would say the smaller ones where there was like double or triple digits of affected consumers could be like tens of thousands of dollars because you have to hire, in most cases, a forensic investigator to do the kind of work that we do.

Um, determine what the actual, what actual data was stolen. Like, let’s say there was a hacking or something and then get notice to those people. And it could be incredibly expensive. It could be like a business killing problem. If, uh, just like one simple hacking. Um, and in fact, like a lot of these weren’t actual hacking, but like somebody left their laptop at a Starbucks and it wasn’t password protected.

And so you just, under the law, like have to deem that as if they had been hacked. Um, if they have sensitive personal information on the computer, so you know what we recommend to most businesses nowadays is you got to get cyber liability coverage, especially if you’re doing anything tech related or e commerce related, where you’re gathering personal information of consumers, but even if you’re not right, like even if you’re just a solely brick and mortar organization, there could be ways where this can can lead to harm because even in just a total brick and mortar organization, you may have sensitive information that Falls within this and again, tens of thousands of dollars for a small business can be the end of the business, right?

Um, and so, yeah, I mean, that’s that’s the way that we’ve generally seen people protect themselves. In terms of GDPR and CCPA, I’d say, you know, GDPR is by far more, um, aggressive and robust in terms of the regulatory element of it than CCPA is, and I don’t want to get into too many specifics here, I’ve spoken on that other times, but basically they require you, among other things, to have privacy policies on your website or other points where you’re collecting customer or consumer data.

Um, And you have to grant rights to people whose data you put to you. You know, the term of art is process like the right in GDPR is famous for being, I think, the first to have a right of erasure. So, you know, someone, if someone had, if you have someone’s data as a business, they can request that you erase it, make corrections to it.

There’s all kinds of rights that are involved. GDPR has enforcement processes and so on and so forth. So the point of this is just is not to get into a specific nuances of each of these privacy laws, but to basically say these laws exist. They’re paying the ass to deal with if you’re a regulated business and many businesses are in Europe, basically all businesses are.

And so you want to make sure that if you’re using AI, you’re not running afoul of these privacy laws, because what happens when you use a tool like, let’s say, Claude. Uh, is your disclosing information? It’s an input, right? You’re putting information into it and you’re getting an output out of it, right? So if you were to say, put in information that had your customers social security numbers in it, uh, that might be a breach of various laws, right?

So the first place businesses need to look, I think, is their own privacy policies, right? What do they tell consumers about what they’re going to do with it. Customer data, right? If you say we don’t disclose it to any person or organization for any purposes, yet you’re providing it, uh, to chat GPT or something where it could be accessed by other people, or even just by the system itself, that might be a breach of your own privacy policy, irrespective of whether a state law, a state’s law allows it, right?

And so I would think that’s the first place somebody should look. Um, the next thing I would say to look at is the system itself’s privacy policy. And I want to read today a little bit from the anthropic policy. Um, this is actually the the terms and conditions, right? So they have, they have privacy policy and then a separate terms and conditions of use, um, the, let me find this part here for a second.

In the terms and conditions, they talk about outputs and inputs and who owns what, um, and and use of materials. So what they say is this, they define materials as prompts and outputs. So that means the things you put into it and the things you get back out of it. And they say we may we, uh, and this is anthropic quad may use materials to provide, maintain and improve the services and develop other products and services.

Uh, and then they go on to say that they won’t train their models on any materials that are not publicly publicly available, except in 2 limited circumstances, which are actually quite broad. The 1st is if you provide feedback. Of course, they can use your feedback to train their materials, or their models, rather.

And the second is that if someone’s models are flagged for trust and safety review, they may analyze or use those materials, uh, to improve and detect and, you know, model their policies on. And I had seen this In the next section, but it says that feedback includes things like rating and output in response to a prompt like by using the thumbs up or thumbs down button.

And so you know the folks at Anthropic might agree or disagree with me, but like let’s say you put in some information. They give you some output and you like it or interacted in some way that could be considered feedback and they could use that information to train their models or for other purposes of developing subsequent software systems and so.

If you’re inputting, you know, protected customer personal information, personal data, however you want to define it, and it’s somehow being used by these models, well, then you’re giving that data to somebody else, which might actually be a violation of law. So I think companies need to be incredibly careful with this.

You know, Clause is actually, uh, the anthropic policy is actually, I think, better than the open AI policy, which. Uh, I think expressly says that they just they can use inputs or outputs. I’d have to go back and find that. So don’t quote me on it. But yeah, at the end of the day, I mean, I’d be very concerned about that.

I, you know, when looking at privacy policies, a privacy lawyer is going to ask you where, what are the outside sources you’re using to gather this information? And, you know, if you drafted a privacy policy 2 3 years ago, or even when GDPR came into effect, I think it’s imperative that you take a look at it now, given that the law, given that AI exists, right, didn’t exist in 2018.

This stuff didn’t exist, right? And so these are just brand new issues. And I think everybody out there who’s got a business should seriously consider looking at the privacy policy terms and conditions and everything else and make sure that you don’t run afoul of these. Even the privacy law side, I want to turn to confidentiality for a second, which can be much broader, right?

So if you’re dealing in a business to business setting, um, you know, you’re going to probably have NDAs in place. Even if you don’t have NDAs, you’re going to have business to business contracts that have confidentiality provisions in almost all cases. And so the things you disclose, even if it doesn’t run afoul of privacy laws, could just run afoul of these NDAs.

And you know, what I think Is it sort of obvious, but it needs to be said is if you break a law, right? Or if you like violate a privacy regulation, nothing might come of it, right? You know, you have to the regulator has to find out about it and then decide to prosecute you or, you know, issued administrative penalty or, you know, commence some sort of proceeding against you.

But if it’s just a business to business contract and the other side finds out, there’s a much higher likelihood that a civil suit is going to be filed against you or arbitration. So, you know, if a business is using a data that’s subject to a confidentiality agreement, just submitting it to AI could be sensitive pricing information, vendor information that AI is going to take it, possibly train future models on it and maybe spit it out to other people.

I mean, you have to be very concerned about that. And I think the third bucket I wanted to talk about on this on this topic is just regulated professions, right? So we’re both lawyers. We have duties of confidentiality to our clients and some basic prospective clients. And so if we’re using AI, like all this stuff could apply to us and we could be running afoul of things like the attorney client privilege.

Various other doctrines, just our confidentiality requirements under the ethics rules. So we need to be careful. And of course, we’re not the only profession that has this right. So you could be a CPA. You could be whatever. You could be just working in a regulated field where you can’t submit things outside the context of your employment.

And so I think. Most of the bus that are on this call are probably going to have to deal with the confidentiality issue in some respect and potentially privacy issue as we go forward with this, and I think the laws are going to have to evolve to deal with this because right now they’re not, I don’t think they’re not, they’re keeping up in a really effective

[00:35:17] Jonathan Bench: way.

Yeah, it’s interesting. A couple of things you said, uh, made me think about, about two tools that, that people use frequently, and one of them absolutely is Google Translate, at least in, in my world where I’m dealing with contracts across borders, and, uh, a lot of clients will say, well, I, you know, I dumped this into Google Translate, and they, they may not think that, uh, You know, Google’s got its own terms and conditions regarding how that information is going to be used.

Uh, you know, including some of those sensitive inputs. You know, we think something we use in the privacy of our own home, it’s no big deal. Uh, we’re just, it’s just a translation tool. But, um, I haven’t read Google translate the terms and conditions yet, but I’m sure they are, uh, very favorable in terms of what Google can do with that data.

And the other is, you know, if you have, a lot of people might get a contract from their lawyers, a long contract, or from the other side on a transaction. Dump it into ChatGPT or Claude and say, give me a summary of this or what, you know, what are the terms, what are the key terms of this contract and unless they’ve, uh, you know, they’ve got a subscription version or a, a more localized version for them that, that respects the, the, the privacy data inside that they could be exposing.

Like you said, they could be violating, um, General general privacy laws, but also specific to, uh, to that transaction where you and I write those contracts and we make sure that in our NDA is that it’s it’s the company and every person who is with that company, including employees and contractors are also going to be beholden to those, uh, those rules to keep things, uh, you know, to keep it out of the public eye.

And I think using those, uh, using anything like that, any AI tool without knowing where the data is going, certainly, uh, should think twice about that. Well,

[00:36:52] Griffen Thorne: a couple of other points to to follow up on this is like, number one, people might be sort of like, okay, this guy is just sort of ringing a bell.

That’s something that might not be a big concern because privacy who’s going to dump like a customer social security number into. Privacy laws generally apply to, you know, personal information, which is extremely broad, right? So the breach response laws are more narrow, but just these laws like CCPA, GDPR, it’s like basically anything that can be used to link an individual to a data set, right?

So it could be very innocuous things that might not seem like a big deal. On the NDA front, like you noted, It can apply not just to the company itself, but to the employees, agents, whatever, but in an NDA setting, you’ll see specific carve outs for instances in which one or both parties to the NDA can disclose information.

So usually it’s like if you’re subpoenaed by a court or if you’re subpoenaed in connection with a court proceeding. You know, you don’t have to comply with the NDA or if you, uh, you can disclose it to your attorney, the information to your attorneys or financial advisors who are assisting you with the deal or whatever it might be, right?

So they’re limited. I think in the coming years, we’re going to start to see carve outs again for use of AI tools, and that’s going to be heavily negotiated in contracts. You know, I also think, though, that people really do need to be aware who are in regulated professions of this stuff. And I, like you said, just dumping stuff into To some of these programs without properly, um, removing information about the transaction can have pretty dire consequences.

And also, like, just keeping in keeping in mind, we’ll come back to this later that sometimes these programs aren’t exactly, um. I don’t want to say honest because it’s not a person, but I’ve had instances where I’ve tried to use, uh, some of these, uh, AI tools and I’ve gotten information back. That’s clearly just inaccurate, right?

And if you’re not following through with it correctly, that could lead to problems. And like tying that back into privacy for a second, you know, uh, there’s users will have a right to you. Correct to cause you to correct information. So if I somehow made something incorrectly in addition to, you know, potential defamation liability, false advertising liability, all that stuff, you could be facing, um, data subject correction requests.

So a lot of a lot of just burgeoning issues that really remain to be fleshed out in the coming years.

[00:39:24] Jonathan Bench: So would you say that then, and I’m not a data privacy lawyer like you are, so I’m going to ask a dumb question, which is, is, is the key issue with, with data privacy, um, disclosure and consent? Are those the two big things, right?

I mean, it’s most important that you disclose, like, you know, you’re talking about these, um, uh, about these programs, and as long as they’re disclosing what’s happening, and they And they receive consent. Is that is that all a company can do? Is that basically, you know, and then be responsive if something happens down the road?

I mean, is that are those really the two silver bullets to make sure that companies can be as protected as possible in this? You know, in our current age, I mean,

[00:40:01] Griffen Thorne: it’s the last time to be much more complex than that. But those are kind of the bedrock principles, right? Like, you want people to know what, uh, what your what information you have about them when what you’re doing with that information.

Uh, all the laws have various, um. Things that you’re allowed to do with customer information and like the ways in which you’re allowed to do them. So, for example, consent is sort of the baseline. If you have consent to do something, it’s fine, but there’s some cases where you don’t have consent and there may be other justifications for using or disclosing a customer’s or just a person’s personal information, right?

But yeah, I mean, if you have fully disclosed the circumstances to someone and get their informed consent, informed Implies necessarily that you fully disclosed to things like that, then generally things are okay. Obviously, it’s in practice much more complicated, but, you know, if you’re going to use if you’re a business, let’s just put aside like a regulated profession for a second.

But if you’re. Just in a business and you gather customers personal information and you use it in a way that you didn’t disclose in your privacy policy or disclose it to an AI tool. And it’s not been disclosed and you’re potentially going to lead to some backlash. So I think, um, being as candid as possible.

I mean, I often caution people to when they look at a privacy policy that I put together to. Think of any potential way you might use data in the future, even if you’re not doing it presently, because you’re going to need to make sure you’re as broad as possible. And so I just think that, yeah, like you said, disclosure and consent are probably two of the most important concepts in this area.

[00:41:43] Jonathan Bench: So maybe you can talk for a minute about the, uh, and then before we transition to kind of the inputs and outputs on the IP side, which I think is super fascinating, um, we have, uh, 15, 18 minutes left before we’re going to, we’re going to quit. So if anyone has any questions, feel free to drop those in the Q and a, and we will, uh, we will review them and discuss them, uh, toward the end.

So Griffen, what about, uh, you and I kick this around in terms of, uh, You know, copyrights and who owns, who owns the data going in, who owns the data going out. I know you talked about this in terms of Claude’s, uh, Claude’s disclosures, but can you talk really, really kind of generally about, about that? It’s, I’m, I have a cool case in China from last year that I can talk about as well.

Um, it’s just, to me, that’s one of the most, uh, kind of big question mark question, big, big questions I have is, is just how does that work on the inputs and outputs?

[00:42:37] Griffen Thorne: Uh, this could be its own like five part webinar, I think. And when we talk about intellectual property, there’s four different really kinds of intellectual property that people talk about, right?

So there’s trademarks, which is, uh, you know, a word or slogan that’s used in a business setting to convey a source. So think about Chiquita bananas. You’ll see the sticker on a Chiquita banana, and you’ll know That’s the it’s going to taste exactly like every other Chiquita banana. I’ve tasted right? And so that’s the trademark.

Whereas a copyright is generally think about an artistic piece. Software can have copyrights. It’s less functional and more expression. But again, you know, software. Admittedly, I’m not copyright has not been one of the things I’ve historically focused a lot on been more trademark and other things. The third bucket is patents and so patents are issued in order to protect technology.

Some sort of creative and useful invention. So think about. Maybe a new form of computer, right? Or a new form of cell phone and get cell phone embedded into your ear or something, right? That’s going to be subject to a patent and all the components of it could potentially be patented separately, depending on the technology and how it evolves.

Um, that sort of thing is. I mean, I don’t do patent law either. It’s you have to have a science background, which I don’t have, although I’ve done some patent litigation transactions related to patents, but the actual patent law itself can be very technical and complicated. And then the 4th area is trade secret law.

So trade secrets are cool in the sense that you can. They can be the same subject matter as a patent, uh, but they’re not registered with any public agency and they have, unlike a patent, which has protection for, I think, about a 20 year period, a trade secret has protection potentially forever. So long as it’s kept secret, so this is very important.

Right? So, uh, the classic example, everyone uses definitely a cliche, but I’m going to use it again. It’s like the formula for Coca bowl. Nobody knows what the formula for Coca Cola is except like three dudes. Uh, they’ve never disclosed it. It’s never been patented. If it had been patented, the patent would have ran out by now and everybody would be making Coca Cola.

But they’ve kept it under wraps this whole time, uh, despite like heavy efforts, apparently to do corporate espionage. And because of that, they still have the, not only is it, is it only known to them, But if somebody were to sort of reverse engineer it or anything, they could sue them and stop that from happening.

Right. And there’s strong federal law protections now, which came into place in 2016 with the defend trade secrets act, which before it was all state law thing, sort of like, you know, privacy in a sense. Um, for me, most of my career is focused on trade secrets and trademarks and litigation. And then again, in the transactional world, um, a lot of my clients don’t.

Go through the patent process for various reasons, and it’s just easier for them to protect things through trade trade secret protection. And the general ways you do that to keep something secret is first of all, keep it a secret. But if you have to disclose it to outsiders, let’s say your your company created some kind of cool new technology and you want to keep it a secret.

And, uh, someone’s an investor considering making investment into the company, You get an NDA in place from day one before you disclose anything to them, because if you don’t get the NDA in place, right, then you’re jeopardizing the secrecy of it, right? If you get the NDA in place, you have at least that base level of protection.

There’s other things, of course, you have to do to protect the trade secret, but it varies on a case by case basis, you know, walling off the information, not letting people who don’t need it have access to it, like the janitor doesn’t need access to your secret sauce, uh, things like that, right? So how does that fit into AI, right?

If you have a trade secret and you input something related to it into an AI type system that’s not, that says we’re owning the input or you grant us a license to use the input, like, you may have just killed your trade secret, right? And again, trade secret may be the formula to create something, it may be the invention itself, but it can also be like data or information, right, depending on the context.

So you got to think about this in the broadest possible sense. And I want to go back to. The anthropic terms and conditions for a minute where it talks about ownership, right? So, uh, where is it here? Give me a moment.

Might have lost this part of it, but Yeah, so, um,

okay, so this is what this is actually OpenAI’s policy, which under the content section of their terms and use says as between you and OpenAI, to the extent permitted by applicable law, you, which means the person using it, retains ownership to ownership rights and input and own the output. We assign all right title and interest in and to the output.

So, It’s not 100 percent clear in practice like how this would exactly work because OpenAI is saying you own what you input to us and you own what we assign you to the extent permitted by law, our ownership of the output. But then it goes on to say elsewhere that we may use content. To maintain develop and improve our services, blah, blah.

So effectively, I look at this is almost granting open a license to use your content, right? Um, even though it may not say that explicitly in the part I’m reading. And so if it comes to trade secrets, even though you have. you retain ownership of it, you’re disclosing it to someone who might go around and use it or might not be under a confidentiality obligation to the same level, you know, your NDA partners would be.

So I think that’s a big risk. Um, and in general, you know, each of these Programs is going to have just completely different terms about what, who owns input, who owns output, who can, how they can use it. Like I read earlier, Anthropic, and in some cases, if you provide feedback, which is loosely defined as potentially even a thumbs up or thumbs down, like you use it to model its software.

So, I, in effect, I think of that as almost like a license, even if you retain ownership to use it. And so then if you own it, you’re granting a license, you know, license to a third party. You have to think like, is this, is this even exclusive? And I may be being a little extreme here, but these are some of the technical things we need to work through that really depend on the type of AI system you’re using on a day to day basis.

[00:49:16] Jonathan Bench: And that’s interesting, because I think in the, you know, you and I are working through a big transaction right now, a big international M& A transaction and the idea that, um, you know, the seller of the asset, seller of the business is going to have to make significant representations and warranties, uh, uh, including around IP and, uh, we’re going to see those creep up more and more, uh, you know, we’re going to have to pull the sellers and find out if they’ve been using, if they’ve disclosed any of the IP, any of the trade secrets into, uh, Uh, you know, into an, an AI, uh, that could potentially have, have voided or at least corrupted, uh, you know, devalued a bit some of the trade secrets or, or other IP.

I mean, it’s, it’s really, um, far, far reaching. And, and those, those M& A agreements with their very robust terms and conditions, very robust reps and warranties, um, those, those encompass everything. I mean, we write, we write it over and over again. And, uh, and those are the things that sellers, uh, sellers can be on the hook for, certainly, and buyers should, should understand that there are risks associated with IP disclosures, uh, around these AI tools, and there will be, I mean, we’re at, we’re at the very early cusp of how these are going to be impacted in our, in our transaction documents.

[00:50:25] Griffen Thorne: Well, I have two predictions on that front. I mean, because I, when I started working on M& A deals five or six years ago, uh, and that M& A just means buying and selling businesses, um, I never really like you see representations and for some context. I mean, just the seller is saying, Hey, look, these I promise you that we don’t have any tax debt, except that we, you know, that might be listed out here.

I promise you we don’t have any lawsuits going on, right? That’s a representation. Or if they do, they say, I promise. These are the only problems. I didn’t see those kinds of things about data privacy initially, but now I see him in almost every deal. There’s it. You know, a page or two of data privacy representations.

So, yes, I expect that there’s going to be all this stuff in AI. The other thing I think is that companies are going to start to adopt policies about allowable AI technologies, right? So, I think, like, um, if you work in a company where, you know, right now it could be Claude, it could be Bart, it could be ChatGPT, whatever, uh, you’ll start to see companies say, hey, we only want you to use If you’re an employee of our company, we only are going to allow you to use bar or whatever it might be in accordance with your, uh, your services for the company, because that’s like the privacy policy we think is the best, right?

That’s the one we think is like the least issue and integrates best with our system. So I think you’re going to start to see things like that happening in the near future. Um, as well. Yeah, and I know Jonathan. I know we only have a short time. So before we’re over, I just want to quickly talk about the marketing side of things, too, because it sort of relates to I P, you know, if you’re using a I to generate, you know, content on a website.

Advertisements this product descriptions, product labels, stuff like that. Um, you need to be concerned about advertising and marketing liability. Uh, federally speaking in the U. S. that will come through the Lanham Act, which is the same law that essentially regulates federal trademark law and. Very big concern right now with AI is false advertising.

You know, under federal law, false advertising claims don’t require like specific intent to do wrong. It’s just like, did you provide a false statement about your product and did it have a tendency to deceive people? And so, you know, if you use AI to generate ads or something and it makes incorrect statements, like your business could be on the hook for that.

Seriously, by either a consumer or a competitor, or in some cases, like a state attorney general consume so you potential issues there. So I would say, like, if you’re using AI for those purposes, people need to review all of the outputs before anything is employed because false advertising liability can be very significant.

And like I said, nobody wants to be sued and. Okay. Personally, I mean, you got, I’m not sure if there’s a lot of lawyers on here, but there were some lawyers who were sanctioned last year, uh, they had used chat GPT and apparently it just made up legal cases that they used in a case brief, like just completely fabricated them.

And they got, it was very embarrassing. I would imagine for them just on all the legal blogs for like months and months getting dragged and they ended up getting sanctioned by a court, which is not something you ever want to have happen. Right?

[00:53:36] Jonathan Bench: Um, yeah. That’s the generative part of the generative AI, right?

Let’s see. Let’s see what’s generated from the data. That’s the

[00:53:43] Griffen Thorne: and I mean, I was I’ve like toyed around with my bio on the website and which I need to update. I’ve plugged it into some AI and then like I noticed it’s saying I have credentials that I’m like, wait a second. That’s not exactly right, you know, and that’s me, right?

So if I’m a business, you need to be very careful about that kind of stuff because obviously. There’s plaintiffs lawyers out there who are just itching to sue businesses on a class action basis for falsely representing something about their product. It’s a lucrative line of work for those attorneys. So just be very, very careful when you’re using these tools because they’re not known to be.

They’re known to give you answers that you want or need to have, but that doesn’t necessarily mean that facts won’t be assumed in a way Thank you. That is bad. And obviously they’re not programmed to lie to you, but these kinds of things tend to happen. I think anybody who’s used these things enough knows that it’s a big risk.

[00:54:38] Jonathan Bench: That’s great. And so I think in terms of my parting thoughts, I want to touch on what the U. S. federal regulators have been talking about on the AI landscape. So the FTC, Federal Trade Commission, governs kind of the general business landscape. And, uh, it, It subpoenaed Microsoft, uh, Alphabet, Amazon, Anthropic, to find out what they’re doing in AI, to find out who they have contracts with.

I mean, they have pretty broad latitude, uh, to find out what’s going on to make sure that the, the, uh, competitive landscape is still level and, and they don’t elbow out all the small and medium sized companies that want to play in the space as well. Um, The FTC has, has said that they, uh, especially don’t want companies to overpromise what their algorithm or AI based tool can deliver.

So these, this is especially pertinent for companies that are going to be using custom AI tools, which are much more easier to, uh, to build. I mean, I, I think that, uh, projects that would have taken 8 to 12 months to build, you can now do in about a week with a good team. And so it’s pretty fast turnaround.

Um, for, for companies that want to deploy their own custom, uh, custom AI. And so definitely keep that in mind that, that whatever, like you said, this dovetails nicely with the advertising. And what are you actually saying your company or your, uh, you know, your service can provide, what can your, what can your, uh, What can your software actually do?

And don’t, uh, don’t overpromise. Um, and that fits into the, into the SEC as well. The, uh, SEC Chairman Gary Gensler has, uh, warned firms as recently as last month in a, in a discussion about AI washing, right? We talked about greenwashing and other kinds of washing that companies were doing. He said, um, that AI is, you know, if you’re raising money from the public, if you’re offering and selling securities, uh, you know, you still have to give full and fair disclosure and then let the investors.

Decide. So you can’t just say we’re an AI company or we, you know, we use AI extensively in our, in our company. And that’s why we’re raising money now. And that’s why we, uh, we’re optimistic about the future of the company. So even that kind of thing, you have to be very careful of. Um, I think this is very interesting, actually, uh, a statement from Gary Gensler on AI.

Uh, and this fits into my blockchain practice as well, Griffen. Because, uh, in the blockchain world, people develop smart contracts and they deploy them out to layer one networks. And then they say, you know, I, I don’t control the contract anymore. And so if you think of this in, in context of blockchain and now AI, as Gary Gensler said, he says, uh, artificial intelligence as we know it now still has humans in the loop.

There are humans that put Uh, that are putting the AI model into place. So there are humans that are still have responsibility for that AI model. So I thought that was interesting. And then the last piece of the Consumer Financial Protection Bureau focused on, of course, on companies have interactions with consumers on the financial side.

Uh, you know, chat bots that could be taking in communications and then giving out information that’s not accurate. That’s especially, um, Especially scary, I think, for companies that might be trying to deploy these technologies, um, on anything to do with consumer finance. So those are kind of the three big, uh, three big federal agencies that, that are on my radar, especially FTC and SEC, though, in terms of what, uh, you know, the competitive landscape and then companies that are trying to raise capital, uh, you know, whether public or private placements of securities.

So I don’t know if you have any, any parting thoughts, Griffen. We, I think we did, uh, we covered a lot. Certainly it was, um, interesting to hear your perspective. Appreciate it. Yeah, thanks.

[00:57:57] Griffen Thorne: You too. And I think, um, well, one thing I didn’t cover was defamation laws. Obviously, if we’re talking about false information, you could also get sued for that.

The other thing is, like, should people be using AI to draft contracts? And I think that the answer is probably not at this stage because, you know, if it’s a very simple contract, like I’m selling you car for 500 on such and such date, I mean, maybe that can be drawn up the right way. But I’ve seen, personally seen contracts that should have been simple.

That even AI fudged and I think the issue is like the more complicated you get the harder harder it is. So I guess if you have the ability in your lawyer to like prompt AI to draft a contract and redraft things and fix it and you know handhold it throughout this whole process. Maybe you can wind up with a good product, but like at that stage you might as well just use a contract and change it that you already have.

I don’t think the technology is there right now and I think that you know it could be there. Maybe. You know, doing legal briefs for courts is, is different because it’s more, um, argument based. But then again, you, you had those lawyers who got sanctioned for the thing, making case sites up. So

[00:59:10] Jonathan Bench: double check citations.

That’s right. Yeah.

[00:59:13] Griffen Thorne: I mean, people are probably like, well, this is just a lawyer saying lawyerly things and they, they, you know, want to keep their job. I mean, there is that degree, obviously, but at the end of the day, the whole point of a contract, someone told me one day early in my career is that it’s not to like, Okay.

Memorialize the terms of the deal. It’s so that when things go wrong, you can persuade the judge of your side, right? That’s the whole point of a contract. And if you’re farming that out to a technology and not like thoroughly reviewing the provisions, or maybe you’re not even someone who’s qualified to review, like, what a detailed indemnity provision means or something, um That could lead to some serious consequences.

And like I said, I’ve seen that happen a couple of times already. So I think maybe the technology will replace us one day, but I don’t think it’s there quite yet.

[00:59:58] Jonathan Bench: Excellent. Thanks. Griffen, appreciate your time. Thanks everyone for tuning in. And we’ll, we will see you on the next webinar. Looking forward to it.

[01:00:05] Griffen Thorne: Take care everyone.

[01:00:06] Jonathan Bench: Thanks.