Video: AI and the Law: Legal Implication & Risks for Marketers | Duration: 3368s | Summary: AI and the Law: Legal Implication & Risks for Marketers

Transcript for "AI and the Law: Legal Implication & Risks for Marketers": Alright. Good afternoon or good morning to some. I'm sure good afternoon, good evening, wherever you are. Thanks so much for joining us today. I'm Adam Kaiser from 6 Sense. Think we've got pretty much everywhere 12:03, so we will go ahead and jump in. This session, make sure you're in the right place. AI and the law, legal implication and risks for marketing. And so before we jump in, just gonna say your things will be recorded. I'll be sharing that later on as well. And I am I am our vice president of growth and brand marketing here at 6 Sense. But I am very excited to be joined by Kara Larson, who has got a very impressive title across our legal team assistant general counsel for privacy and compliance. But Kara, let us let the let the folks know. I was gonna call you my partner in crime, but that seems somewhat Yeah. I don't know if that would pass, legal review. And, Adam, I actually have a surprise for you, an honest to god surprise because we we did not discuss this before. I don't like this. I thought what better way to kick things off on a presentation about AI, risk, and marketing, than by asking some generative AI to give me some jokes to share about this. So I'm gonna share a couple of my favorite ones and favorite is used very loosely here and perhaps ironically. So one is using AI and marketing without checking compliance first is like sending out party invitations and forgetting to book the venue. The party might be fun until the authorities show up. Oh, snap. That's both funny and biting at the same time. When it comes to AI and regulations, it's like teaching a robot the rules of the road, except instead of traffic signs, it's privacy laws, and the penalty for running a red light is a really expensive fine. Wow. So, I think, you know, we're still working on like humor in AI and so that perhaps, you know, connecting both with the consumer and with your brand in an organic way, you know, may or may not be a great use case, for generative AI at this time. But, super, super glad to be here. If I can work in a couple more of these jokes, I absolutely will. The model was obsessed with making jokes about coffee breaks for some inexplicable reason. So I'll, maybe share a couple of those later on. I also think there's probably I'm known for dad jokes. There's a thin line between generative AI jokes and dad jokes probably as well. So that's all I'm looking for. So anyway, Cara, talk let let the folks know again, like, with 6th sense. Like, you're you have a pretty large purview. I think it's sort of it's so relevant for this conversation. Yeah. No. Thanks, Adam. So I do a little bit of everything at 6 Sense. I, work on, you know, protecting our employees' data, but then also our customers' data. Also reviewing, vendors as they come in and assessing risk there. Advising on risk and compliance during product development, during marketing campaigns, website configurations, consent collection, mapping all of our uses of personal data, you know, in the platform, but also in our own, like, use as a controller under the GDPR or as a business under the CCPA. So I'm kind of all things, at all times and in all places. There's very little that isn't under my purview or that I'm not determined to put under my purview in what I hope is a friendly way. Well, I can attest several times I've gone to care with campaigns and messaging me like, 1, is this right? And 2, is this crazy? So 2 very important questions to ask, with that. So, just to give you if you don't know 6th Sense, and we're not gonna make this a 6th Sense commercial as much as we can, but, really, just to give you a quick overview, 6th Sense was founded based on a very simple question of, wouldn't sales and marketing be a whole lot easier if we just knew who was looking to buy our product? Simple question, pretty difficult to answer, but we've aimed to, over the years, develop technology to do just that. So, really, the idea is that 6th Sense can tell you when accounts are in market. So an account that's a good fit for you, but are they also showing behaviors and activities that could indicate that there is a buying process going on, about to begin, in progress, and so forth. Who is on that buying team, who to contact next, how to contact them, and what they care about, and now in the age of generative AI, what to say and even saying it for you. So the the platform has gotten, some legs, there as well. So the first thing that we thought we would do is kinda set the stage with a bit of a poll, and this is just to get a sense of what's going on. So the question is, how are you using AI in general? So hopefully, I hit the right button. I think I did. And I I think you did. Let's, let's get some answers in. While you're getting some answers in, I do have more jokes. Oh, boy. Okay. I mean, go for it. They say data is the new oil, but with all these regulations, it's more like the new gasoline. Handle it carefully or you'll blow up your marketing campaign. I'll be aggressive. I think I'm just laughing because I'm nervous by just how bad the jokes are. I mean, it's just that's what's happening. Okay. It's a how are we doing on this whole? Let's see. I think we're are we done? Alright. Let's let's throw the results out there. Alright. So alright. It's pretty good. We've got sometimes, most of something that's that should probably be sometimes. Actually, let me most of the time, always, rarely. We have no never. So we've got everyone in the room is at least doing something with AI. So that's, but here's here's a here's sort of now a a transition to actually and I'm gonna go ahead and close the poll and then stop sharing. I did. I hit both buttons. Very good. Okay. So guess the question becomes now, let's talk about a little bit of definitions. Right? Because think AI let me get screen share back going in. AI has become continues to change, to evolve, but ultimately let's define it. Kara, I know you guys have a question. Yeah. I love definitions. I love making sure that we're talking about the same things. We have found that AI can very often be used as shorthand for generative AI specifically, But AI, of course, is much broader than just generative AI. AI truly means it's the automation of tasks by a machine or computer that otherwise would typically require human brainpower. Beyond generative AI, there's examples like voice assistants, like Siri or Alexa, trying to trying to cause everyone's devices to go off right now, or recommendation systems, like on Netflix, Spotify, or YouTube, Chatbots, you see that very often in, like, a customer service or, like, an inbound lead situation. And then when we're talking about generative AI specifically, as opposed to AI that helps automate tasks that would otherwise be performed by humans, which is extremely broad. Generative AI is going to be AI that can create novel outputs and content. This is based upon a model, and this is going to be different than providing responses that are exactly as preapproved or making some sort of prediction about a dataset. It is called generative AI because it's actually generating content and output, that can be very often is novel, but of course is not always novel, which we'll get into when we talk a little bit more about risks of using AI. Yeah. This is important distinction again because it's gotten it's gotten a little out of hand too in terms of, like, use you know, especially vendors that are selling products. Everything has AI in it at this point. I think I bought, flour to to bake the other day. It said it was AI powered, which I thought was a little bit, a bit much. But so let's actually go now. So, obviously, with AI, I'm sure, like, especially marketers who are sitting in here probably have concerns. So I am going to launch into our next poll because we are a poll heavy group here. So this is head over to the, pulse tab again. What worries you about your team using AI? And the quick quality control, security and privacy, copyright infringement IP or confidentiality. And we are gonna get into all of that and more. Because it can be I think there's also been a lot of, like, sort of fear mongering out there too. But, you know, a lot of this comes down to, like, be responsible and, like, you wouldn't take your company's product code and staple it on a wall in a public place. So why would you put it into, you know, so we could get into I mean, Adam, like, we we have not met everyone in this world. Never say never is what I found previous life as a criminal defense lawyer. Just some of the unbelievable things that people would do and think were a good idea, like sharing another AI joke. Using AI in marketing is great, but remember, just because AI can predict your customer's favorite pizza doesn't mean it should text them at 2 AM about a sale. Compliance first. I question whether or not yeah. I can predict your customer service with that. But that being said, let's take a look at what we got going on. Alright. Number 1, quality control. Interesting. I may have skewed the results a little bit with my jokes. You yeah. That's true. They're like, this is all about that issue now. It's copyright infringement, IP, security and privacy. Interesting. No confidentiality, but I guess probably security confidentiality. Maybe there's a little bit of, the way that that was worded. So alright. So some interesting, interesting concerns and thoughts out there. So alright. So let's go back here now for a second to sharing again. We're talking about responsible use of AI. I think we're gonna kinda hit on a lot of those concerns over in that poll as well. So, I don't know, Kara, if you wanna jump right into meat and potatoes of data privacy and compliance, and we can sort of riff back on from there a bit. Yeah. Absolutely. And I know that not everyone has the time or inclination, to spend hours and hours reading through the GDPR and the CCPA and its accompanying materials and guidance and legal decisions. There's just a lot out there. In terms of GDPR and CCPA, and of course without providing legal advice because I'm an attorney but I am not your attorney. A couple of the biggest concepts that you want to be aware of in your role as CMO or as part of a marketing team is going to be these ideas of transparency, fairness, and consent. So transparency is are you telling people what you're doing with their personal data? Of course, personal data is going to be data that relates to an identified or identifiable natural person, if we're borrowing from the GDPR there. Are you telling people what's going on with their personal data and what personal data you're using? That's going to be transparency. Fairness is going to come down to, well, is it a reasonable expectation of a person that their data would be used in this way? Essentially, does it pass the smell test? And then, of course, you're going to want to think through consent. Are you in a jurisdiction where consent is required? If you are, what does that consent look like? Is it an opt in consent that you're looking for like a GDPR or with the CCPA? Are you in a situation where you need to provide an opportunity to opt out? Now with GDPR, although there's the EU AI Act that really directly addresses the use of AI in the EU, the GDPR already had provisions related to making predictions about a person through the use of AI. Keeping in mind that the GDPR was passed in 2018. In a lot of ways, it was a little ahead of the curve in thinking through the use of technology and how it interacts with the human experience, and whether or not using technology and removing humans from the mix results in an untoward legal impact on a person. If you're using AI to make a prediction about a person that could or does have legal effects like eligibility for a job, creditworthiness, those types of things, then you're looking at you do need, again, without providing legal advice, I would strongly consider using consent in such a scenario. Those are your baseline principles to consider when you're thinking about AI. What are you telling people? How are you telling them that? What type of permission do you need to get? But in order to make sure that you are fulfilling your transparency and your fairness principles, you also need to have your own understanding of what's going on. What personal data is being used? How is it being used? Where is it going? Who's storing it for how long? These are all considerations that should be thought through beyond just, hey, we're using your AI, is that cool? Check the box, yes or no. And then, you know, Adam, I don't know if you wanna take the 2nd bullet related to, like, bias and fairness in AI, but it kind of flows directly from that personal data aspect. Yeah. I mean, it's it's interesting. Right? There was some tests done, and this is this goes back a ways now. I mean, obviously, AI hasn't been this generative AI movement movement or I should say movement. The world evidence is pretty new. But even on some of the image creation that's out there as well, right, which you've got, like, content creation at the text level, like, that was sort of the first thing. You're seeing image creation now, video creation now, all that type of thing. But there was an example shared of, you know, create an image of a successful CEO of a Fortune 500 company. And every time, it was producing a picture of a, you know, 50 year old adult or adult, of course. I don't have kids at CEOs. Anyway, adult male, generally why so the the bias was inherent in the in the models themselves. Right? So as you're going through and, again, a lot of this is just if you're generating content, it's in your marketing campaigns. However you're using it, like, just looking at it from just, like, you're human. Right? Don't we're not gonna ever rely on AI to produce and go, at least in terms of and we'll talk a little bit about how could AI be actually acting upon you in marketing campaigns and so forth. But you're not gonna create that. No one's gonna just create that image and go or create that content and go. Someone's gonna take a look at it. So it's it's it's artificial intelligence, but some of these inherent biases that sometimes exist in humanity have made their way into the models as well. So it really it creates, things you have to be careful of whether or not using AI, which is kind of an an interesting twist on that as well. And I think, also, when you think about just AI generated content in general, and I won't go down to the legal part with this because it'll be a little bit slippery, but things that are interesting there, like, if if if any of you have played with the Adobe suite. Right? So Photoshop has become, and actually pretty much the entire suite, fully AI aware. Adobe has their own model, for doing generative on imagery. And early on, there was questions of, okay. Well, who actually owns this? Because it was pulling in stock and video, and I think contributors to stock and video were able to say, yes. This can be used in ad generated content. Early in those betas, it wasn't allowed for commercial use, these images that were being created, at least by licensing terms. Now Adobe has a model for charging for credits and so forth, but it could be a slippery slope, especially with, like, what's training your model and the output? So I don't care like, again, I don't wanna talk intellectual property law because that in my world. But curious if you think about how do you walk that line and still be responsible? You know, you're using AI to help you, but you're not violate just straight up violating some IP property like IP rights of of company or individuals. Yeah. That's a great question. And and it it goes also back to your point on bias and fairness in AI and, like, having this human review element. To the same extent that, you know, you may or may not use, you know, an agency to put together your content for a campaign. But it would be very rare that you would go to the agency and say, hey. We want a campaign focusing on this segment. We want it to be bright and uplifting. You're good. Just, like, let us know after the campaign runs and and how it does. You wouldn't use AI in that same way either. And I think it's, like, the fundamental question that you should ask before using AI, you know, specifically when we're talking about creating marketing content because that is where you're going to see the most risk when it comes to IP, is, like, would you would you run the same play or the same process if there were humans involved? And if the answer is no, like, no, I would probably wanna take a look at the content and give some sort of approval. You should probably do the same in AI. AI helps build in efficiencies. For example, you've got a pretty good idea of the creatives that you want to run, and the primary image that you want to use. But maybe you're playing around with the background, and so you're generating different ideas for, well, you know, like, I want a gradient from red to blue. Well, I want, you know, maybe a beach scene. Those types of things, great efficiencies there. But the more you start handing over more and more responsibility to AI, especially if you're wholly removing humans from the mix, that's when you're gonna start to run into more issues. Not because you intended to do anything, but just because, like, you know, we've seen how sometimes, especially with especially with generative AI, the prompts can be manipulated in a way. You know, it's usually by some sort of, like, red testing or even, bad actors or someone looking for clout to show like, hey, look, even though it's not supposed to make Mario anymore with the image generation, I can still make Mario if I say Italian plumber. But it is something to be mindful of In the same way that you would run your typical compliance checks to make sure that you weren't violating any IP rights of a third party before you push a campaign live, you would do the same thing here with with use of AI. That's going to be the best way to make sure that you are staying, as buttoned up as as you as you can possibly be in in this area. Again, not legal advice. Work with your compliance team, but work with them in the same way that you would for any other campaign. And I know that we're getting to transparency and explainability. This time it's gonna be because of, decisions made by the AI. Like, how did the AI reach that decision as opposed to, hey. We wanna use your personal data. Here's what we're using it for. Now you've put the information into AI. There's been an output, and either decisions been made, a recommendation has been made, or content has been generated. Can you explain how you got there, and are you transparent about, like, how you got there? I'm just kinda zipping along because I know we're getting Yeah. There's one thing. Yeah. I think there's one thing on transparency and the explainability part. Like, if you look at a lot of these models now, like, CHA2PT in and of itself now, like, as they release more models, the models themselves are not are now showing you the thought process that goes into it, which is good for just understanding where the how they got to the answer, especially on, like, scientific and sort of, like, higher level deeper order thinking. But it's giving you the explainability there too. So it's become part of a lot of these offerings to say, like, hey. Well, this is how we got there versus what when these things started, it was, like, a prompt or when it was magic, and you kinda didn't know what was going on. So I think that's especially, I think it's the latest the latest chat gpt model. It's got really, really thought out view of that as well. Yes. I was really pleased when I saw that because I think it both helps with compliance obligations, but it also helps for end users. If you're trying to figure out a better prompt, you're like, Oh, I see what happened here. I was trying to get it to tell jokes for use in a presentation about AI and marketing, and compliance, and not like separate jokes about AI marketing and compliance. Ethical considerations and corporate responsibility. Apologies. I've got some earbuds that just will not stay in. In terms of ethical considerations and corporate responsibility, important to keep in mind that this is a quickly evolving area. Regardless of the ultimate risk versus reward analysis being performed, there might be huge reward in using AI. But if your company positioning is such that it's extremely risk averse and not risk tolerant, It may not be a great fit at this particular time because there are still wildly varying degrees of comfort just at a very high level of corporate viewpoint and thinking through. Then of course, corporate responsibility. Keep in mind that especially with the use of generative AI, less so with other forms of AI, you are talking about an environmental cost. That will need to be factored into, again without providing legal advice, your greenhouse gas emissions calculations to the extent that you are doing those and have made commitments. It costs power and power is emissions. That is something to keep in mind as well. That's a good one. Let's go into AI marketing use cases. So this is What? This is the this is good the fun stuff. Right? This is where I can sound smarter. So, again, I think when you think about generative AI, obviously, the first thing that comes to probably most people's minds is just pure content creation. Right? There's, you know, there's so many tools that continue to pop up, but you can go and, like, hey. Write me a blog post about x y z, and and you will get an output in a lot of cases that is you know, it needs some work, but it gets you started. My team here at 6 Sense, we we dove into using generative AI pretty early on for the on the content side. About a year and a half ago, we made a purchase of a, you know, one of the more commercially available tools. And it's become this you know, we haven't had to grow the team, but our output continues to grow up. But it's it's actually incredibly powerful and more powerful, I think, for ideation and figuring out what you wanna cover, the types of content you wanna write, giving you that sort of initial outline. But we you know, and we will we've trained it on our tone and our voice. And we had one particular project where we acquired a company, and there were about a 100 blog posts that we wanted to keep from their blog that were the content was good, but it was not on brand, voice, and so forth. And we were able to use AI to do kind of a a rewrite in our voice and tone, and it saved us an obscene amount of time. So, like, that that use case on content creation, like, that's pretty locked. And I think you see so many ways you can do it. There are tools that are doing it on the fly inside of, like, I mean, my Grammarly, which I used to use to say if I was writing in passive voice when I'll do generate it all kind of built in. So, like, that's pretty straightforward and and I think really, really helpful, from a content perspective, b to c, b you know, b to b and so forth. Then you're getting into, you know, understanding, like, predictive analytics and lead scoring. Again, like and this isn't necessarily all generative AI, but what I think what this sort of openness of using these tools is made people realize just how much data can be analyzed so quickly and used to apply to, like, understanding leads, where they are in their journey, and so forth. And the 6th Sense, again, not putting in the commercial, but we've had a predictive, product for 10 years. And it's all based on understanding a huge dataset of, like, here's all the stuff that accounts and opportunities that you've worked in the past have done that tell us, okay, well, we know what good looks like. We know that these are the things that all the accounts that you've won have done. And, you know, again, there's millions of signals. And then be able to take that, analyze it, and say, okay. Well, guess what? Here's 5 accounts over here that are showing similar activities. You might wanna spend some time over there. And then using that to also to to score it and so forth as well. So that opens up huge possibilities for sales and marketing efficiency, where to focus your time, kind of all those, great things. You know, also and I talked a little bit about this earlier, the visual creation of photos and and and videos now is is that's really kind of one of the things that's happening. The the videos that are being created now, it's actually scary. Right? Because you can generate a person walking down the street and you'd be like, was that shot with a cam? Was that real or was that AI? So, again, like, that's helping with any number of different things. Right? Maybe you're tired of using the same stock photo in one of your websites or or in an ad or so forth, but now you can start creating entire collections of imagery that's gonna have your brand colors, kind of follow your overall aesthetic, but they're they're original because they're being generated off of, a back of AI. So that can be a huge a huge thing as well. And, Adam, I wanted to actually pull a question, that was asked directly relevant. Larry asked, you know, what are your thoughts on using AI photography tools to create a person's biography photo versus like a traditional photographer? I've got some thoughts. I'm not sure if if you have some thoughts. I do, but I think you actually have you've actually done that before. Right? You've used those tools? Yes. Yeah. I have. I have. So, first, like, leaving judgment at the door. There are a number of reasons that a person may want to use generative AI headshot technology, beyond cost. There are matters of time and effort in terms of scheduling time with a photographer, finding a photographer, carving out time in your day. My current LinkedIn picture is from 13 years ago. I just have not had the time. That said, I think it is important to support the people who are performing the work. Photographers serve a very important role and function, and the quality is just not going to be as good as if you go to a photographer. You are getting a lesser product if you're using generative AI to produce a headshot. But in terms of personal data and consent where you're creating your own headshot, you know, are the controller and decision maker when it comes to your personal data, you know, without providing legal advice, it would behoove a person to double check the terms to see how their image is going to be used. Is that image going to be used to train the model? Are you okay with that if it is? Or if you're not okay with that, is there another tool that you can use, that would provide further assistance? So I I think it comes down to, you know, your priorities, your time, and, you know, the ultimate, like, end products that you're looking for. But again, like, the quality is just never gonna be as good as a actual human photographer, at least as of right now. Yeah. And I think they're the tools are getting better in a sense. Like, I remember when they first came out, like, I think I ran one of mine through it. I'm like, that doesn't even remotely look like me. And, like, look, I want that sort of, like, chiseled superhero face, but I'm like, alright. I know. I don't have that. But they're really kinda far off. And I think they're getting better and better. So it's also, like, beyond all those really important considerations, like, do you is it being somewhat truthful in what you're trying to represent? Like, I think that's kind of an important question for yourself. Again, of course, where you're using it comes into play. So, it took me 7 years and 94 photographers to get that headshot that I like. That's for a different day. There. Those are all, good points. So I know we're coming up on time to get to q and a. So, chatbots, again, I'm sure we've all, at some point, have interacted with a chatbot. What's what's really happening now is whether it's customer service, if it's, you know, on a marketing website, there's a lot of questions that you don't need a human for. Very transactional business questions, very basic support and customer service questions. And these things are getting, you know, more and more advanced all the time. There are products on there right on in the world right now where you go to a website, you could ask the thing a 100 questions, get all your product questions, and, like, you know, sorry. I'm ready to talk to buying to a salesperson to to buy it. So that's becoming, pretty much standard and really coming down to how you train that model. And I think all through this whole kind of the entire conversation around generative specifically, it's only as good as the data you give it. Sure. The models have, you know, their base knowledge of the world and all that, but if you're not training it on not only your knowledge, your company knowledge, you know, your SME knowledge, but your brand tone of voice and all those things, then it's it's still I don't wanna say it's dumb, but it's somewhat generic. Right? So, like, that that's something if you're gonna start to deploy these things into the world, another use case could be, there are now products that will go and prospect, send emails, and and take back responses from potential customers. Again, if they're not trained, those models are not trained on what you sound like and your knowledge, again, it's gonna come off as generic, probably silly in some cases. So, really, in any of these use cases, in any of the tools, and we'll talk about sourcing a little bit too, but understand how you can infuse these models and these products with your content and your data and your knowledge to really make the AI talk, look, and speak like you do. So alright. Now we're gonna get to sourcing. This is you know? Also, just your in the the docs tab in Goldcast, we've actually got a vendor purchase checklist that you can grab. But I am gonna throw this one to Kara. I've gone through procurement processes myself on AI, tools, but, Kara, I'll just throw this to kinda start. Again, actually, I'll just say clear business objectives, pretty clear. Right? Why are you actually going for an AI tool? What are you trying to accomplish? What's the outcome you're looking for? Whether you're buying AI or anything else, you're gonna, hopefully, you're gonna do that on any purchase. But then we get into privacy data security, so I'll throw it back to you, Kara. Maybe you have another joke about sourcing. I don't know if you do a procurement jokes might be another area we could go into. I mean so balancing AI marketing and compliance is like trying to juggle chainsaws. Dangerous if you don't know what you're doing, but impressive when you get it right and no one gets hurt. So that much not better. Okay. Yeah. Yeah. Yeah. So we're, you know, we're kind of at the same like, the good news is, The good and bad news is that the quality is pretty consistent here. But with sourcing, which I'm just thrilled to talk about, I know everyone here is sourcing and they're like, oh my God, I can't wait. This is my absolute favorite topic. I love everything about the procurement process. But fundamentally, and like as Adam said, you love procurement? It came off mute. I have a t shirt actually, I thought where it says I heart procurement. But I'd say our procurement team at 6¢, it's good people. They're good people. Yeah. Like Adam was saying, the biggest thing that you can know is what are you doing and why are you doing it? Because when you have a very clear vision of this is the tool that we're trying to get, this is what it does, this is our outcome, it helps your compliance and legal teams really critically look at the terms of the agreement to say, okay, you want the model to, you know, you want this AI tool to be trained on the tone of your marketing copy and go back and tweak existing copy or blog posts so they sound more aligned with the tone. We see that this agreement says that in terms of the data that's being uploaded, there's some user information that's being uploaded. But wait, what is this? It says that it wants our entire Salesforce, all of our CRM records and a few from our map. That doesn't really track. Or the output here is making changes to our IP, like our blog posts. And so therefore we do want to own the output because the intended use case is we want it to adjust the tone of some of our pre existing blog posts. But this agreement says that we actually don't own the output of the model. That could definitely be a problem. In terms of what challenges you could run into when using AI tools to collect and understand customer data, that is going to really depend on a couple of things. The first is what type of personal data is being processed, specifically from what geographic regions, what are your requirements in those regions in terms of transparency, fairness, and notice, and consent? But also, like, who are your customers? Do they have a heads up that you have AI in your services or that you plan on using AI? Is your customer's expectation that you would tell them about this use of their data? That goes beyond just the contractual language, but in terms of being a good steward of your customer's data, do they have this expectation that you would inform them of this use case? But that also goes back to this prioritization of data privacy and security. If you are going to use an AI tool to collect and understand customer data, the controls in terms of how that data gets used and whether or not it can be used to train the model and how it gets stored and how it gets retained become absolutely critical. Because presumably, you are going to tell your customers, hey, we're onboarding this new vendor. It's going to be great. Let us know if you have any questions. Their questions are going to be, how are you protecting our data from being used in a way that we don't want it to? You do need to know information from that vendor about what are their data handling practices. That also goes back to their transparency and ethics. In terms of transparency, at a baseline, and this goes back to explainability, the vendor should be able to say broadly, this is the type of model that we're looking at. Like is it a classification model? Is it a large language model? Is there generative AI? Are we running just a basic logistic regression? Being able to articulate at a high level the type of model will also help your compliance teams and yourself understand if there's something like hinky going on with either the promises about what the services can deliver or how the data is intended to be used when being processed by those services. That all rolls up, of course, to compliance with legal standards. Depending on the use case, depending on the sensitivity of the data that's being processed, the legal standards may vary and also depending on industry. If you are a banking institution in Germany, you have a different set of legal standards than you do for a small business in Louisiana. Some regions and jurisdictions have AI specific rules. Colorado just passed an AI Act. There's the EU AI Act. But knowing exactly the type of AI, the data being submitted, and the purpose of that AI are going to help your compliance team make those compliance determinations about how can we use this tool in a way that is not going to run a foul of regulatory requirements. I'm going to kick it to Adam to talk a little bit about usability and integration. Yeah. I know this is I mean, there's so much there to to talk about on the on the legal side, but, you know, understanding, 1, how you can use the tool with other parts of your tech stack. Right? So in terms of, it's about convenience. Right? So if it if it's a if it's a writing tool, where can I use it to generate content? You know, something as simple as, like, you know, I work a lot in Google Docs. I wanna generate content in there, or I work in Word or something like that. You know, on the design side, is it is it in the applications I use every day that may not have AI features, but there's there's some plug ins or some other technology to kind of help augment the use of that? So, like, that really comes down to, you know, you wanna use AI as a productivity gain. You wanna use AI as something to make your teams more effective. Maybe it's something that your teams aren't as good at and they wanna streamline a process so they can get better at it. So thinking about don't think about the AI tool as just this thing at the side, but how can it be a part of these other technologies, again, without having security and content and and other issues would happen to the data? And one of the things that you'll probably see, especially if you're in a larger company, procurement teams and security teams are and I can I can attest to this from just experiences at 6¢? We now have processes around, hey. If you're gonna go and buy something that is inclusive of generative AI, they've gone through and they said, okay. Here's the different types of data and kind of risk modeled. Right? So if you're doing x, right, it's publicly published data, nothing else, it's it's green. You can go, we're not gonna put a lot of scrutiny on it, All the way through to, like, are you nuts? So I think there's a lot of work that goes on with compliance teams, procurement teams, info sec teams to help marketers, whoever else are sourcing these products, to do it in a way, like, you don't need to become an expert in this stuff. Right? It's more about understanding some of the inherent risks, understanding what you're trying to achieve as a marketer, but then letting legal and procurement do the do their thing, so they can help you accomplish your goal without creating a mess, a risk, you know, that type of thing. So and, of course, the reputation of the vendor. You know, kinda again, just like you'd buy anything else. These last things support, product, all those things are are completely independent of it being, an AI tool, but just good things that you would wanna do, when you're sourcing from a third party vendor for a tech tool of any kind. So, Karen, if any before we go to q and a, any other thoughts on sourcing or overall legal compliance? I I think if there's one lesson to take away from today beyond, you know, here are some, use cases for AI, which do run the gamut from generating content, but also, you know, slicing and dicing some of your existing data. The biggest, like, takeaway is knowing what's going on with data. Like and that applies to basically all aspects of a of a marketing journey. Right? Like, you should be able to articulate, here are the types of data that we're collecting for this campaign. You're measuring, you know, the performance of the campaign anyway. And, you know, here are the plays that we're gonna run. Here's the content that's gonna be used. But more deeply than that, when it comes to the tech that you're using, how does it work? You know, what data is being collected? And if you can squarely and confidently answer those questions, you're going to get through procurement much more easily and smoothly, and also be able to source vendors in an easier way because you're going to have a really clear idea of your objectives. You're going to know what's going on. I I think that's probably going to be the biggest takeaway is like knowing what's going on. Yeah. And I love what you said about being stewards of the data. Right? I mean, again, to my earlier point, you don't have to be an expert. You've got teams that are just going to support you. But you also can't wash your hands and be like, I don't know. I don't know where the data goes. Like, make sure you like, again, you have to have a cursory understanding of what's happening. So you're responsible within the organization, and you gain the confidence in your working with the teams that you're with as well. So that said, right on time. We have some great questions already that have been asked. Let me You want me to read them? You know what I mean? Yeah. I'm just figuring stuff out on the fly. Jason asked, what risk do you encounter when you contract with third party vendors who unknowingly use AI on your behalf? What level of due diligence is necessary or appropriate? This is going to come down to corporate risk tolerance. You only know what you know and you don't know what you don't know. In terms of risk, if there is AI and specifically with personal data, you may have a notification obligation, depending on where you are in the chain, you know, if you're collecting that personal data, how you're using it, what this tool is. Not knowing whether or not AI is being used could put you in a situation where you're unable to make, required disclosures that AI is being used, in this manner. Now the level of due diligence that's necessary or appropriate is going to come down to, you know, your corporate risk tolerance, your past relationship with the vendor, and like how far you want to push things. If you've got a good relationship with the vendor, you can always just ask. Say, hey, how does this work? Is AI being used? If you don't have time, don't have a good relationship, or just still have some lingering questions, you can kick it to your legal team and say, hey, I I don't know if AI is being used here or not. Can we, you know, maybe send them some sort of contractual update, that specifically, like, clarifies obligations with respect to AI? Can we send them you know, can I kick it to security? Can security send them, like, a risk questionnaire? Maybe someone more technical on your team, like or, you know, within the company can come in and get with their engineers or their, support individuals to kind of get more of that information. Because it is really important to know at least broadly, you know, what's going on and and whether AI is being used or not. Really good question. Nice. Alright. How did you do that? I feel like I don't the one at the left, it says q and I with a little person. Yeah. And then you hit share. I don't oh, there it is. There it is. Okay. Teamwork makes the Teamwork. Alright. How do you see the rise of AI impacting legal standards for marketing? Are there specific changes to expect in the next few years? I actually have a thought on that, but I'm not a lawyer, so I probably shouldn't answer it. Or maybe No. Why don't you go go for it? I don't think it makes any difference because I think marketing should have a a standard. There's a legal standard. Right? You know, I just think about in your day in and day out. You probably know maybe you don't. But you think about, like, what has to go to legal. Right? If you're you're you're let's just say you're writing to a blog on a on a couple of weekly basis once or twice a week. You're probably not creating content, though. Every blog post needs full on legal review. Right? It's editorial content. You know, there's a process you follow. Maybe you have an organization that requires some cursory look versus rolling out a large scale campaign that could and I I I could pull up a specific example of a campaign that we did earlier this year that I talked to care about in terms of, like, how we were mentioning popular culture and kinda and it was clear this is a large campaign that needed legal oversight because it was so expansive and it was covering off a lot of different things. Do those standards change on what the test is for going to legal or getting legal involvement in marketing based on AI or not? I don't think they do. I think you have a high standard that you maintain. I can be completely off base, but that's kinda my thought. No. I I I think I I think I agree with you broadly. I think about it in 2 in 2 buckets in terms of, like, anticipated, legal standards and and changes. One is, to the extent we see additional obligations to identify the use of AI, like in the ad creative itself, like image generated by AI or, in connection with the use of a chatbot, you know, under California's Bot Act. There may be an obligation to make additional disclosures around your use of AI. That's going to be probably the biggest impact on marketing is like, oh, how do I fit this in, and where do I fit this in? On the other side of things and this is this is I've been thinking about this question since I saw it in the chat earlier is, the ability for plaintiff's attorneys, you know, who are filing test cases, who have new legal theories, to more quickly identify potential claims because they are using AI to either crawl websites really quickly to look for missing mandated disclosures and then find a sample plaintiff to bring a cause of action, or because they're running, you know, maybe they've already got the company that you're at, like in their sites and they're running like AI analysis on your site to figure out if all of the disclosures are there. Or mapping personal data from dropping a cookie in one place to an ad being shown in another place. These types of large data analysis are ripe for the use of AI. I think even though it doesn't change the legal standard for marketing, I do think we start to see some slight uptick in risk because of how it becomes easier to try to find these claims. That's very good point. That's why I'm not the lawyer. Alright. So let's try another. If the content is created through AI, how does the content need, if at all, to be cited to recognize that you may not necessarily be clear on sources? So there's actually one before we get to the legal part of this, there's one interesting thing too in kind of in the in the AI world. If you haven't heard the acronym RAG right? So RAG is basically and I it's retrieval augmented generation. Right? So if you think about large language models that sort of exist now, you ask it a question, you ask it to create content, it does it based on what it's been trained on. You know, like, chat cpt has been trained on the Internet of the world, you know, whatever it may be. Rag now are these are these solutions where, yes, you have a large language model, but you can then introduce to it large other documents, other pieces of content that you could be using to help aid that training and help answer those questions. If you're for example, you could take, like, a company's 10 k report and put it in and start asking questions about it and getting information about it. In my mind, I think, like, a lot of these tools I've seen it where if it's specifically pulling content from a source that you've given or trained on, it's likely gonna cite it. But I guess I guess the question to you, Kara, as well and this is, I think, become I don't wanna say it's a slippery slope, but see more and more companies are like, hey. This is AI generated. Just to sort of say it's AI generated. Do you think it goes further than that either now or in the future? Yeah. Potentially. So I think there's a there's a couple of moving concepts and moving targets. There's identification of AI generated, either content or conclusions like, hey, your credit score is x. These were the factors. Here's how they were weighted. That's gonna be explaining how that credit score was reached. Now with something like generative AI, that it that might be more of an identification. Like, AI was used to generate this language. But to, for example, like OpenAI and its own efforts around explainability, now does offer like how it arrived at the conclusion. Now, a description or explanation of how you arrived at a conclusion, of course, is not the same as saying, these were the specific sources of data used to reach this conclusion. It's just saying, here's how we got here. I think as we see more guidance and regulation, there's going to be an additional clarity about the situations in which, like, you're okay to say, you know, generated by AI or here's how it was generated by AI, or here are the specific data sources or elements that were used to create this, you know, piece of AI generated content. It is still a little bit murky right now in terms of what to do when. I would not cite as a source like, if I was writing a, like, law review article or magazine article or something like that, like, I don't think I would be at the point of, like, citing information gleaned from a generative AI model unless the model itself pointed me to verifiable sources that I could you know, I've used it as a jumping off point to find some potential sources. Very good question. It's a very interesting question and one that I think we'll see further development around. I'll say there's, like, entirely new, like, fields within law, within IP law, even all these things that are popping up in specialties and all that too. Right? So that'll be really interesting to see. Okay. Always an adventure. Is that did we get them all? We might have gotten them all. I think there's only one. Yeah. What happens I'll share it. What happens if an AI if a company's AI generated content accidentally infringes on someone else's intellectual property? The the lawyer answer is that it depends. You know, it depends on whether or not, accidentally to me is like unintentional. Of course, like, if this is something that happens in a vacuum that there's an infringement of IP rights and no one notices, then the answer is nothing happens. If there's an infringement of IP rights and someone notices, you're looking at anything from a takedown notice to like institution of legal proceedings, to the extent that there's been a deprecation in value of those IP rights because of that infringement. That'll come down to the rights holder and what they determine is the best course of action. In any event, your best way to avoid finding out the answer to that question is to run your creatives and anything else through compliance review before pushing them live. Makes sense. Well, with that, we're coming to an end. Do you have a closing joke that you would like to share? Yes. Yes. I do. I asked AI to help with marketing, and now it's trying to sell me a smarter version of myself. Well, I think with that, we will say thank you all very much for joining. Cara, I am thank you for bringing those jokes. I think it it it brought the game in this presentation up probably 4 or 5 notches. So really just remarkable. And now I think I will rip that off for all webinars going forward. So perhaps I will not be doing many webinars going forward after that. But Oh, no. Me neither. I'm definitely not gonna be invited anywhere after this. But it was good while it lasted. It was. Well, thanks everybody for joining us. Appreciate it. And, have a great rest of your day. Thanks, everyone.