Welcome to the Marketing AI SparkCast, where AI meets modern marketing. Hosted by Aby Varma, a global leader in marketing, AI, and innovation, each episode brings you real conversations with leaders putting AI to work in strategy and execution. This episode is brought to you by Spark Novus and the Marketing AI Pulse community. Here's your host, Aby Varma. Hello, everybody, and welcome to the October episode of the Marketing AI Pulse monthly brief podcast. This is where we discuss stories relevant to CMOs and marketing leaders at the intersection of marketing and AI. This episode is sponsored by the tenth Marketing AI Pulse. This is happening on November 12 from 01:30 to 05:30PM at Atlanta Tech Park in Atlanta. We will have speakers from Numerail, Jabra, to Zapier, along with agency leaders and practitioners. Expect panel discussions, breakout sessions, and plenty of time to connect with other marketers. And you can learn more about it at marketingaipulse.com, and you will find a link to register. I would love to see you guys there. Alright. And with that, I am going to welcome my friend and cohost, Matt Cyr. Matt, good to see you, my friend. Aby, how are you? How were how are things in the land last week? It was it was really good. I just got back from MAICON, which is the marketing conference in Cleveland, which is what Matt is referring to. Had a really good time. One of my favorite parts of going to that event is really seeing old friends and meeting new people and hearing how everybody's approaching this mad, wild world of AI that is happening all around us. Didn't cut about a huge thank you to Paul Roetzer, Mike Kaput, Cathy McPhillips, and the entire SmarterX team. They really put together an amazing event, and I appreciate the part of the effort that goes into bringing that community together. Matt, I know you were not at the event. Did we miss you, my friend? Yeah. Thank you. I've been the last couple of years this year. I couldn't couldn't pull it off, but I I enjoyed from a distance seeing everybody's post. Got a little bit of FOMO in the mix, but glad you had a good time and not surprised. It's always a great event. Yeah. Absolutely. And to our listeners, Matt and me did not do September episode. There was a lot going on, so we decided to take a break, but excited to dig into this episode. So we have a lot of cool stories to talk about. We got a couple of four things. So we're gonna cover personal AI agents and how they are moving into everybody's lives and marketing teams. Matt is going to also talk about personal AI agents, but in the context of privacy and governance. So really interesting perspective. Then I'm gonna cover something that I hear a lot. I've heard at MAICON a lot, which is CMOs are under pressure from the CEOs and CFOs that, hey. You're... You guys are using AI now. Great. Let's just reduce headcount and cut cost. And, I'm here to tell you there is more than one way of looking at that. And then finally, Matt is going to bring in a home about AI in ads. There is a lot going on as the political climate here in The United States heats up. So excited to dig in, Matt. We have a pack packed episode today. Got a lot of good stuff to cover, my friend. Ready to get going. Awesome. So I'm gonna just get started here. So in late September, OpenAI introduced Dots. So what are Dots? So they are always on agents with their own cloud computer and connected apps. A roll... Rollout has begun for eligible pro, business premium, and enterprise users. But those unfamiliar with personal AI agents, they're really assistants that remember your preferences and carry out ongoing work across connected tools, obviously, with permissions that you provide them. And OpenAI isn't the only shop in town, so to speak, that is doing it. Right? Like, September, Meta introduced Muse, and that works across apps and websites and tasks in the background. Late in October last week, I believe, they announced new models like Muse, Wonder, and additional models that are gonna power some of these personal agents. In August, Grok Bot was launched by the team at Grok in in mid May at Google and our seminar at Spark. So there's a whole host of these agents that are infiltrating our workspaces, and these are ubiquitous, always on, and they're doing things without a specific trigger. So they're monitoring things to try to understand some of the goals and objectives that you're trying to achieve, and then they will help you achieve that. And that could be just alerting you, or that could be doing work. It could be a combination, again, depending on what sources that you are connecting these to, what's the permission, and what exactly do you want it to do. So from a marketing standpoint, it changes things in big ways. Obviously, there is a personal element to it, but then there's also agent based orchestration implications of that. So from a from a marketing standpoint, that could mean delegation. Right? Handing off recurring preparation and coordination with clear instructions. Obviously, you want some human control, but that is a very common use case. Well, go pull a status report, pull some regular data, put some analytics together, come up with the first draft, like some recurring task is sort of a common use case. The second aspect is continuity. Right? So assignment can carry forward between meetings rather than starting over with a new prong. Right? So the marketer still needs to explain what matters, resolve exceptions, and decide what happens next. But that makes supervision, judgment, and direction a larger part of a job as these routine tasks become easier to delegate. So an example could be, hey, before Monday's campaign meeting, gather approved performance reports, audit updates, compare results with targets, and prepare briefing with source links and many missing information flagged. And that could be an automatic part of it. Obviously, you gotta kinda tie it into all the sources where this where this data lives. But really bringing all that information to you in some automated way where you don't have to proactively prompt these agents. Then what makes it useful is that clear assignments, direct instructions, schedule, expectation of outcomes. It's a great tool for monitoring. So if you wanna do competitive monitoring, again, a great opportunity to do that. Obviously, the success of that depends on the granularity with which you give it instructions. I just don't say that, hey, monitor my competitors. You could say that, hey, monitor competitor websites. What is the pricing or offers or new developments? It can be very narrowly. You you could scope it in a very granular and narrow way to kinda give you the exact piece of information that you want. Yeah. So so a lot of opportunity. But with such recurring responsibility, the question is, would you delegate that to a personal agent today? And, obviously, you want human... You want the human in the loop and human approval. But, Matt, what is your take on this? Would you would you delegate this recurring responsibility to a personal agent in insistent approvals? Just wanna get a sense of what you're thinking here. Yeah. For sure. It's a great question. I think finally, agents are getting to the point where, these questions have become really relevant. So the short answer to your question is, yes. And there are things that I'm already using agents for that I'm allowing them to do and introduce some delegation. They are pretty low risk things, though. And so one example is pre meeting preparation, allowing agents to pull performance reports and competitor information from approved sources into a briefing with gaps flagged. It's a high volume, low risk task that I think people can comfortably get to understand how these agents work without putting data at risk. But I think anything that leaves the building with my name on it or any of my clients' names on it is gonna have to get signed off by a human. Right? That's really where that shared responsibility comes in. The line that I would draw is that the agent can read and draft, but this person has to decide and act, and that's what keeping a human in the loop really means from my perspective. And the approval really is only half of the question here. The other half is what the agent can see in the first place, and that's something that I'm gonna talk about in my first segment here just to Awesome. Yeah. I think that's a very practical way of looking at it where you wanna make sure that you're using it to help you, but with boundaries at the end of the day. Right? And for individuals that are using it, but I'm also seeing marketing teams begin to experiment with it and tie in cross functional teamwork flows where you have agents that are communicating with other agents. Right? And so what is an approach for CMOs? How should they kinda go about this? So my recommendation would be to choose one one workflow. Start with one example if your team is not using it. It could be, like I said, a campaign preparation, competitive monitoring, customer feedback, essentially, but establish the time that this is required. What is a good... What is good look like? What are the data sources and those things? Very important to Matt's point is set the authority and the ownership. Specify what the agent can read, draft, and change. And this is very important because a lot of times, people give carte blanche access for the agent to do things, and not sure that that is a best practice. So drafting an email, for example, or sending an email may require a different permission set. So making sure that there is a complete understanding of that, and then making sure at the end of the day, a human is accountable and responsible. So make sure you... Whatever workflow that you choose, there is a defined human owner for review and exceptions, and earn the side of caution where you're telling your personal agent where if you don't know stuff or you don't... You're not entirely sure, check with this person or interact with this person on the team. The other thing is the success of these agents, these personal agents, is kind of very heavily dependent on what what is the data sources that they have access to. So in a work setting, you have to make sure, check with your compliance and IT folks or your marketing operations people where... What are approved data sources. And if agents are connected to it, what are... It could be a platform, but you could only have a portion of that platform that is really permitted to be connected to an agent and where administrative controls. And in terms of logging and getting a very good understanding of what these agents are doing behind the scenes is a critical part of making sure that the the agent activity is transparent to you guys. So decide what happens to access on ongoing work. Also, when an employee that is involved, that person leaves, what are you gonna do next? So making sure some of those things are in place. And at the end of the day, making sure that there is a clear line of delegation, source checking, and escalation, right, which includes setup review and corrections and all of those things. Don't get too enamored by the by the speed and the agility of what these AI agents can do and ignore the the accuracy of these agents. So that's sort of a critical part of the equation. So I think in general, if you have not started playing with this, and your organization has any of these tools that have these personal agents, start getting a team's hands dirty. Get your own hands dirty. If you don't have it in a corporate setting, try it on your personal accounts, and get a feel for how these things work. They can be really powerful resources that can really improve your productivity in a big way, and really help you, and be a true sidekick in a really powerful way. But start with what is responsible. Like, start with something useful, establish clear boundaries, and then expand the agent's role as it's performing, quote, unquote, earnings up. Then that is a a a critical way of going about it. Now there are various aspects of agents, and privacy is a big piece of it, which leads us right into, Matt, your storyline. Talk to us a little bit about how you were thinking about privacy and personal agents in an era when agents are live and perpetually on and ubiquitous. So what what could go wrong with Red? So much could go wrong, Aby. You used a really important phrase there at the end, establishing clear boundaries for these agents, and I think that's probably the single most important thing that I would share with people here today. I think for the last maybe two or three, maybe two years, we've been talking a lot about AI agents, and the the fact is that they've been more promised than reality. I think that has changed at this point. I think now with the release of Muse and Dots and Grok and Spark and many others, the idea of agents is now real. And so now I think it's the time when we really have to be thinking about how these agents work, what makes them successful, and what do we as marketers and as human beings want them to be able to do on our behalf. So the context here, and you mentioned it earlier, Meta released Muse in September, became the number one app very, very quickly, passed 5,000,000 downloads in just over three weeks, has a million daily users. And if you've used Muse, I have not yet, but I've used other types of agents, it asks for all sorts of stuff, to your calendar, to your emails, your passwords, your contacts, and not least of which your bank accounts. The internal instructions for Muse, in particular, show that it is creating what it calls a dossier every hour on the user and every one that that user mentions. Those dossiers are mapping relationships between people, goals that users have not stated out loud, really anything that is going on in your communications either just by yourself or with others, Muse and other agents are capturing that information and creating connected dossier of who you are and what you... Who you're connected to and what you're trying to accomplish. So there's a lot that is happening, and as some of the critics of this have said, the interface is designed to encourage you to share as much as possible. There was a Washington University law professor who said that he wants agents treated as, in his words, an ultra hazardous product with strict liabilities, and the market is already reacting here. Schwab is blocking Muse from moving client money. JPMorgan is barring outside agents for real client work, and Amazon has actually blocked Muse from making purchases on the Amazon site. So the fact that these massive companies who have all embraced AI in their own ways are becoming concerned about these agents, in particular about Muse, I think, should be a signal to people that there's a lot here that's happening. And I don't wanna be a fearmonger, but there's also a fair bit that can go wrong. So I think that guardrails idea that you were talking about is really important. So for marketers, what does this mean? I think first and foremost, this means that your inbox is not only yours anymore. Right? If you are allowing these agents into your inbox, it means you're giving them access to things like client information, strategies that you've created, NDAs that you have with clients and others, media plans, vendor pricing. Imagine anything that you might think of as sensitive in your client relationships or personal relationships. These agents might have access to those. So an employee who connects a personal agent to a work account has shared all of that information. No one at any of those client agencies... Excuse me, client organizations has ever agreed to that. So we've we've talked for a long time about what you put into LLMs is really important because if you don't wanna see things appearing in LLM responses somewhere, then you have to be very cautious about what goes in. I would take that doubly so when it comes to agents, because there really is a spidering that happens in terms of how these agents run through all of your materials. The second thing that marketers need to understand here is that consent does not carry over. Right? A user can agree for themselves, but not for the client, colleague, or customer whose emails are getting profiled by these systems. Be cognizant of things like GDPR and CCPA. They will start to pay close attention to how people are using these agents and what kind of information is being shared. So it does put you at a higher degree of risk and potential for difficulty if you're allowing these to have free rein. And finally, in this area, I would say that Asians are creating a new attack service. Malwarebytes, which is a large company focused on data privacy and security, warns that phishing emails can now hide instructions for the AI reading your inbox, so you would not have any sense that this phishing is happening because it would essentially be happening between agents. So that agent that can act on your behalf can also be tricked into acting on your behalf. So a lot to really be cognizant of as we move into this brave new world here. So, Aby, you you just talked a fair bit about the case for handling, handing work to agents. In the teams that you talk to, are people checky what kind of data is going in to those agents before they connect them, or is it not yet top of mind for them? I think it's all over the place, to be honest. Right? Because I think with personal use cases, I think people are much more open to agents just connecting to their Gmails. And let's face it, it makes life easier for them. They can be like, hey, could you remind me of the deadline for this personal activity that was... Somebody sent me an email for three weeks ago. Yeah. You can go in and look for an email, but you can just ask your agent to go find that information for you, and it will, and they can back to reply and reply back and confirm. So there's lots of things they can do. So these are obviously designed to make life easier. But in personal use cases, I see that there are people more amenable to that. In corporate use cases, it is dependent on the size of the organization. So the big enterprise guys have it pretty locked down where these agents are not activated. And any agents, Any ubiquitous always on personal agents are very specifically designed and intentionally kinda made available to the teams. That said, the smaller organizations, again, because they are small, they're dealing with bandwidth issues on those days is when they're really leveraging these agents when the agents connected to their to their CRM and their project management system and their emails and their notetakers, and now they can triage work and be like, hey, what what were the list of actions that Matt and me discussed in the last meeting? And can you please go log it into the project management system? And those are things that are doable today. Right? Like, where it can go Do that. You can add a a review cycle in the middle. Hey, tell me what those actions are. After I approve, then log in into my project management system, and send Matt a follow-up email saying that these are things you can do today. Right? Like, and it's super easy. So I see the smaller orgs are more open to that. And then I've seen hybrids where they will... Organizations would wanna use this, but they have walled off access. So you have your main mail inbox, but then you have a standalone inbox where you let these agents connect to, and it's intentional for the user to drop things in the mailbox that the AI agent has access to. So then it's intentional. So your AI agent cannot see a full mailbox. It can only see what is in the explicit mailbox that you've given access to. So it adds an element of manual work, and it also adds the whole governance part of it. So I think it's a different strokes for different folks in a lot of ways, but I think there needs to be just a lot of education that needs to happen for the risks that some of which you mentioned. Right? Like, which is not common practice. People don't really understand that. People get really mesmerized by those cutesy little characters, the the Dots and the Muse of the bots that these companies have come up with and don't really... Or not fully aware or appreciate the kind of the risk that these teams carry. Yeah. It's it's totally true. I've heard... I just actually wrote a blog post about these cutesy little characters as you called them and what they... What they're trying to get you not to pay attention to. So... Yeah. And I really like the idea that you just described of having stopping points along the way where you are giving the okay for these systems to do things on your behalf. I think that's I think that's really great. So to wrap up this topic, I just wanna share a few things that CMOs can be thinking about when it comes to the use of agents. First and foremost is know what kind of data and whose data is in the blast radius as it were. Right? Before anybody connects an agent to a work account, list whose information it would reach, whether that's clients, customers, or colleagues. If they wouldn't agree to it, neither should you. Right? So just imagine or assume that these agents can essentially get access to anything in your system unless you tell them not to. If you're working with vendors, ask them three simple questions. Is training on by default? If so, how do I turn off? Who else's data gets stored or touched in this process? And then can you prove deletion? If I need to delete something, can we prove that it's actually been deleted? So there's a lot that we can still do here and should be doing to make sure that this stuff is protected. And the last thing I would say here is just sounds a little bit strange, but go into this with a with a lens of distrust. Treat everything that an agent does with that lens. Right? That human approval needs to be in place before that agent acts because they will do as much for more than you think they will because they are capable of doing so. So important topic and certainly something to your point that requires more education and training. Yeah. And the way I think about this, Matt, like, let's take the... Let's take cloud computing. Right? Like, when cloud computing was a thing, there was so much you would cry about from CIOs and CTOs about data living in the cloud and that's not being secure and all those things. And now that it's become the de facto standard. Right? Because it's just more cost effective. So I think who is to say what the future holds? But I don't rule out a time in the future where things like this is just gonna become the norm. This is how work is just going to get done where you have your AI agent scanning all parts of your life, work in Bristol, and helping you along. And and I think institutionally, and in terms of legislation, I think as a society, a lot of things would need to be done to make sure that people really trust it. But it's really hard because it does make life easier, and people are willing to sacrifice some of that privacy and governance to make their lives better. Yeah. That's a great point. And as I'm gonna talk about in my in my last topic. Right? We're already not doing a great job of disclosing when AI is being used for advertising, let alone when AI agents are being used to complete tasks. So really poor topic. Glad we covered it from a couple different directions, but time to move on. You have a... An interesting topic today, which is around the freeing up of marketing capacity. Tell us a little bit about it and how we should be thinking about it. Yeah. Absolutely. So perfect use case, the previous story, right, where we were talking about, yeah, agents, and that that is a perfect example of how it may free up time. And I mentioned I was at the MAICON conference last week, and there was obviously a lot of discussions happening about personal agents and educational and literacy and empowerment, and agentic... The agentic revolution, and all those things, AGI, plus a discussion happening. But in... Surprisingly, in speaking to a lot of CMOs, a common theme kept coming up, which is CFOs and CEOs going to CMOs and saying that, hey. Now that you guys are using AI, can we cut your head count? And I started thinking about that, and there's some data that we can look at where you realize that it is certainly not the only option. That is one of the options, and it comes with pros and cons with cost cutting and headcount reduction. But Workday, this this is a study from early October where they did some research in finding competing expectations for AI and staffing. This is not a marketing only study. So extrapolating some of that data here in the context of marketing, but really, 40% of business leaders expect AI to help them get more from existing employees. 28% expected to reduce headcount. And it made me it made me think about the fact that, hey, at the end of the day, AI is going to free up capacity. Right? So something as simple as that where the existing technology is getting AI already, and that's gonna be a huge time saver for Teams. So what happens to that capacities? How do CMOs really deal with it? The... Like I said, the obvious thing is to reduce cost. But, really, the way I look at it is CMOs have the option broadly in thinking about it in three separate ways. Right? So save, improve, and expand. So from a save standpoint, that's where cost reduction comes into play. Right? That's where you reduce your operating cost. You start looking at bringing agency work in house or reduce retainers or evaluate fewer backfills and smaller teams, and those things come in into the equation. Right? Agency decisions should follow the work, like routine production may move in house, or for agency work, when there's T and M type of work happening, it's just taking a much, less amount of time for the agency or your freelancers, contractors, consultants to start delivering things. So these are all options for a headcount risk reduction. Now it's a common... It comes with risks. Obviously, you're just looking at one side of the equation where you're simply thinking about, like, hey, I'm gonna reduce costs and not really thinking about the impact. You've done a lot of work with agencies now, and you are working with agencies right now. Simply, just because you have AI getting rid of your getting rid of your agency is certainly not a good approach without really doing some value analysis that the agencies were delivering and what you can bring in house. And also, some people don't entirely appreciate the fact that just because they're using AI, it doesn't... There is efficiencies, but those efficiencies are not a 100%. There is still a learning curve, and there's still ongoing learning curve in making sure that there's... Could be a a workflow, a process for your organization, and there's human in the loop involved, and all the take. So just because you have AI that could be certainly efficiencies, but it doesn't necessarily include full on processes. So really carefully analyzing it. And then from a headcount reduction standpoint, it comes along with implications of morale and all those things with the commit team that kinda remains. So in general, I feel the reduce the cost of delivering marketing, aka the save approach, while it is one of the approaches tends to be the most common default way that CFOs and CEOs will approach CMOs with. But there is also two other ways for CMOs to think on how they can really kinda deal with that capacity, and the second one of that is really improve. So make existing marketing more effective. So invest time in customer understanding, could be stronger creative, better targeting, and testing. So, for example, capacity free free for asset production work to test better messaging with conversion as a measure, not just activity, but really the outcome. So name the performance problem first, and then if that is weak, take the excess capacity and invest in improving the... Your marketing effectiveness in that area. And then you finally expand, and that's where you're pursuing new opportunities the team couldn't previously support. That could be test a new segment or take... Add a new campaign or localize a campaign or other things in your... In a typical CMO's marketing backlog. Believe me, there is no shortage of pipeline of work for the average CMO. So making sure that you are taking all that extra capacity and adding it to doing things that the team could not previously support. So an example could be the team that can adapt and improve campaign materials more efficiently, use of that capacity to test an un... Underserved or an unserved segment, which is something that they were not doing before. And, obviously, you need to align with sales and all those things, but these are important opportunities, the expand and the expand bucket that the CMO has. And the right mix, these are not mutually exclusive opportunities. It really depends on what does that mix look like. So as a CMO, you should really look at the impact, the staff reduction. What does that really mean? Like, don't just cut your staff. What does that mean for the remaining work, and making sure that it is truly a staff reduction where it's not really impacting work, or measure the whole workflow. Like, post drafts may create additional fact checking, editing, and coordination overhead. So they may be gains through AI, but they may also be additional work that AI is adding kind of to the mix. So the CMO's job is to really evaluate those trade offs and really understand what exactly is happening, but don't default to cost cutting. Right? And, Matt, so with a... From your standpoint, when a COO is being asked to reduce agency spend or headcount, what would make a convincing case for reinvesting some of that in capacity or growth? I know you work with agencies a lot, but what are what are your thoughts there? Yeah. Good question. To me, I think the upside needs to be as concrete as the cut. Right? Savings number is a certain number, but the gross... The growth story typically is not. And I think CMOs can be better at really measuring and determining what that upside is going to look like. Pick a problem, attach a test, an owner, and a measurement to it, and make sure that you can see where the value came from. Separating production from judgment, I think, is really important. We just... We all know that AI speeds up production, but strategy, taste, client, customer relationships are what you lose if you cut people and agency partners. Right? So I think that trust in those relationships are probably the hardest thing to rebuild. Using real evidence on the work that's remaining, I think, is incredibly important. Right? I think very often, we stumble into the cuts without thinking about the true value of what we are doing instead. And I think really measuring that, I think, would be really critical. Again, putting in place a test plan and and having some data that will show value. And then the last thing, right, from my perspective, cutting is a one time win. Reinvesting is how you show the CFO that AI is a growth story, not just a cost story, and I think that's a real shift that will probably start to take on even greater rely... Or greater importance in 2027. So I would start getting ahead of it now. So, yeah, that's how I would look at it, and I'm sure you have some thoughts on what CMO should do next. Yeah. Yeah. And I think a lot of the cutting is a one time win is where from from a CMO standpoint, there is, like I said, three options, save, improve, and expand. And part of the CMO's role is to really bring their CFO and their CEO into the fold and make sure that there is a firm understanding of the fact that there is more than one option. Right? And also just purely from a savings standpoint, you gotta verify that gain. Like, measure the one workflow that is getting impacted or the workflows that are getting impacted and make sure that the quality holds. That just because you were cutting people out using AI is not resulting to the same or better quality of output in speed, but then also quality. Right? But I think overall, my recommendation for CMOs is that that free capacity is really to be used in a mix. It's not a it's not a linear choice. You can save, improve, and expand in certain areas, and you should do do a little bit of experimentation in these three areas and see... And measure them and see what those numbers tell, what story those numbers tell, and then make sure that you are putting those in front of your CFO or CEO based on your findings where you can you can be like, hey. Yes. As a result of that, we can cut we can cut 10% of the budget, but here's the negative impact. But if we kept that budget, the value of improved conversions, the value of pursuing a new campaign, a new market can exceed the the 10% cut in cost. It can add more to the top line. So really making sure that when AI frees up capacity, deciding where is gonna create the most value, and then changing the plan, the budget, and the assignments in alignment with that. Don't just default to cost cutting and layoffs. So that would be my advice to to CMOs. Alright, Matt. Onto our last story, which is AI in advertising. There's a lot of that happening right now, especially given the election fever, lots of political noise. Talk to us a little bit about what is going on, and what does that mean? Lessons from that whole political talk to us a little bit about what is going on and how marketers can learn from what is happening in that world of politics. Yeah. It's interesting. Right? Because I think politics and all the political advertising is... I wouldn't use the word leading because I'm not sure a lot of it's very positive. But these political ads are, I think, really putting into stark relief the decisions that marketers need to make when it comes to the use of AI. And it extends to really all different kinds of things, certainly politics, but really any brand that you can imagine. So, you know, disclosure about AI usage is becoming a brand decision, not just a legal one. Some brands, and I'll talk about a couple here in just a moment, are labeling the fact that they're using AI on national TV. In general, political advertising does not seem to be doing that. And so the way that this is going out into the world is really setting expectations for consumers of all kinds, not just on the political side. But I'm gonna start there because it's probably the most eye opening example. So in 2024, AI political ad spending was just under $50,000. Now here two years later for this cycle is a $137,000,000 spent on AI and political ads, which is a stunning increase as the technology has accelerated in that time. But to go from under 50 k to a 137,000,000 is really wild. And fewer than one in three of those political ads is saying that they are using AI in them. In fact, 220 ads were evaluated across 34 states. Only 27% disclosed AI usage. I live here in South Carolina, and I can tell you that there are AI created ads all over the place, including of actual candidates. These deepfakes of real politicians, and there are no disclosures, any of those ads that AI has been used. And in fact, 30 over 50 states... Only 30 of the 50 states require disclosure. So a significant portion of the country is is being fanned ads without clarity that these are created by AI. Some brands, though, are going in the other direction. If you've been watching sports in the last couple of weeks as I always do, There are two good examples. Baerskin, b a b r skin, is an outdoor clothing brand. I was watching one of their ads a week or two ago and thought, man, this looks AI. And at the end of that advertisement, it said, in fact, that the entire thing was generated by AI. And then this last weekend watching some football games, there were ads for Huggies. Some of you may have seen this. There was a a spot about premature infants. And at the end of the ad, there's an image of a premature baby in the NICU, and the baby has their fist up, and they've got a boxing glove on the fist. And they make very clear in that advertisement that that was created by AI specifically because they didn't want to have babies in the queue be bothered by being filmed. So in both cases, significant and clear disclosure that AI was used, but on the other hand, advertising coming from politicians simply not happening at all from a disclosure perspective. So from a marketing perspective, this is really starting to create a trust gap. Right? 82% of ad executives think that younger consumers feel good about AI ads, but only 45% of those consumers actually do. And more than 30% of consumers in every age group say that AI ads make them less likely to choose a brand. In fact, what they have found is a disclosure about AI usage does not hurt the sales. 73% of Gen Z and millennials say clear disclosure of AI use would raise or at least not change their likelihood to buy something from a brand. And, honestly, disclosure at this point is the cheapest way to earn trust from a consumer today. But I think especially given the amount that these politicians and their packs are spending on these ads, that the political ads are really setting the reference point here. Every undisclosed deepfake ad is teaching viewers to doubt what they're seeing on the screen, and that includes your ads. So as you are making your own advertising, it's really important to make a decision about what you're doing, how you're doing it, and what you're disclosing. But I think AI has become such a ubiquitous part of the mix for these CMOs. Again, I think that disclosure standard has to be in place before you really need one. You don't want somebody calling you out. IAB has a framework that's a good starting point, labeling AI that could mislead about what's real, such as synthetic people, cloaked voices, depicted events. I think that kind of disclosure will go an awfully long way. Treating disclosure as part of the story, I think, is really interesting and probably not something that's gonna need to be done for a long time, but I think that needs to be done today. Baerskin and Huggies are good examples. Explain why you use the AI, not just that you did. And then the last thing that I would say for CMOs is plan for regulation. The EU's AI Act has labeling rules that took effect in August. Any campaign that runs in Europe is gonna need to disclose. I think we're probably going to find all sorts of other standards being brought into place. And so better to get ahead of it and build that trust with your audience now than be called out for doing something that people recognize and that you haven't agreed to. So a really interesting time, I think, here for marketers. I know it's something you think a lot about. Any any final thoughts on the idea of disclosure here, Aby, before we wrap it up? In my mind, there is no other way than just to do it. Like, it's getting mandated, I think, earlier in the year. Was in one of our podcast episodes where New York state actually has a disclosure law where you gotta conspicuously disclose any AI avidars used or people who are fake used in your advertising, and there are several other states that are that are working on it. So to me, I I don't think it's it's a matter of if you should do it or not. I think it's just a matter of when it's gonna be mandated. And outside of the mandate and the legal part of it, like you said, at the end of the day, it's about brand trust. So you wanna make sure that people are trusting what they're seeing, especially if your brand name is associated with it. So key part of it. Matt, thank you very much for your research and your stories. Always enjoy talking through these stories with you. Me too. These are great. This is good stuff and super timely. Yes. To everybody listening, thank you for spending time with Matt and me. We'd love to hear what you're trying and learning within your own teams. You can connect with Matt and me on LinkedIn. If you are attending the tenth Marketing AI Pulse on November 12, Matt and me will both be there in person. And if you're not part of the Marketing AI Pulse community, you can check it out at marketingaipulse.com. It's free to join, and you will really connect up with an amazing community. Till next time, I hope you guys have a great time, and have a good one, man. You too. Good to see you. Thanks for listening. Big thanks to Spark Novus and the Marketing AI Pulse community. You can join the community at marketingaipulse.com.
