Welcome And What We Cover

Chris Norton 0:00

This morning's event is called Protecting Your Brand in the Age of AI Misinformation. And I'm hoping that you find you guys get a lot out of this because we have done a few workshops with clients on uh AI crisis management. We also we used to do a lot of stuff around social media crisis management. Will and I do quite a bit of media training too. So today's event sort of combines all that, but we also got some entertaining videos in here and some really useful stuff and a process that Will's gonna talk about um later. Right, so yeah, as I said, this is protecting your brand in the age of misinformation. Um, and what we're gonna cover in the next 60 minutes. Uh, we've got a couple of slides on us, but that's it. Don't worry just about who we are. Um, so you who Prohibition is, um, why AI misinformation matters in 2026, why brands uh uh can lose control so quickly. Um, I mean it was quick in social media, but in the era of AI, it's it's bonkers. Just look at how many blooming AI tools are coming out. They released five ChatGPT 5.3, not last week, and the day after they released ChatGPT 5.4. Uh so yeah, it's a it's a bit of an AI space race. So it's good to just stay on top of these things. We've got some case studies and some industry learnings. We've got plenty of entertaining case studies as well, so you get to laugh at other people's misfortunes or not. Um, and then wait when I uh when AI hits, so how to fight back, which Will's going to cover, and sort of the first 24 hours. So um my name's Chris Norton, I'm the innovation lead. Uh I'm massively into AI, working on several different bespoke tools, um, both for clients and for ourselves. Um, and obviously it's the area of technology, technology and uh strategy is where I love to be. So um, but I'm the founder of the agency, and yeah, that's me. So, Will?

Speaker 4 1:56

Yeah, hi everyone. I'm Will. I um I tend to look after creative and strategy in the agency. I also do a lot of um crisis management training, and what I'm gonna talk about in the second half of this is how um crisis plans, which never used to take into consideration social media, now need to also take into consideration um AI. So essentially your crisis plan needs to evolve. And we're very much of the view that we need to be proactive when it comes to misinformation, not reactive. Definitely a hot, hot topic, isn't it, Chris? And one we're having many more conversations about with our clients um every week.

Chris Norton 2:32

Yeah, we we did we uh Will and I hosted a uh bespoke um AI session with a client and on misinformation and crisis, and we they they had the the brand will have to remain nameless, but it's a global brand, and they shared from around the globe some several different examples of things that had happened to them, and then we picked a few bespoke ones that we created as well. So we're doing workshops on this sorts of stuff already. Oh, Will, you've just shared that poll. Sorry.

Speaker 4 2:58

I've immediately pressed the wrong button. I was just gonna say I'm gonna be publishing various polls as we go throughout this. So by all means, um ask questions in the chat, and if you can let us know your um your views in the polls as well. I'll be then looking at the results at the halfway point.

Chris Norton 3:13

Yeah, so um yeah, we we're but we're Prohibition, we help um ambitious brands grow faster. So we work with everyone from Oxford University, Barbour, Goldsmiths Jewellers, Rolex, um Fentimans, Yorkshire Water. We're based in Lee Leeds in the UK. Um the agency we've got 28 consultants, but we do everything. Um we I'm from a PR background, but tech and PR, which is why I've always been into social media and AI and all these different things uh that come from tech. Love that area. But we're a consumer uh and B2B uh agency with sort of 60% consumer and 40% B2B. I've got a B2B team. Um the different things that we offer, um, in terms of services, we offer um what we call a growth catalyst, which is sort of demand generation campaigns, so we help people to get um better uh traction in it in a category if that's what they're after. Um we also do, which is kind of where this area is Reputation Guardians. So we're proactive and reactive. We can help a client with crisis management, either online, offline, whatever that is. Um we've been involved in tons of those. Um we also do what's called a launch ignition. So if you've got a product launch or you've got to launch a service or something, we can help you do that in terms of video, um, print, broadcast, whatever that is, uh basically to help a client take away the hassle of launching a new product. And then the final service that we do is like a behavioral shift. So we work with quite a few charities as well and uh in education, so uh because they're two of our areas of expertise as well, and and that we like to help out, we've got a reduced rate for charities, but um yeah, we like to help shift behavior where where possible, so it's like influencing we've put here mid medium-term attitudes, but you can basically do all sorts at scale, and we help a client to do that. So let's get into the good stuff then. So, why does AI misinformation matter in 2026? Well, Will, you need to mute, mate, because uh I can hear all your typing. Um, yeah, AI misinformation, it erodes customer trust, um, it creates product confusion. So if you've got a product, a great example was yeah, if somebody does it does a fake video of your product, and we've I've seen examples of that already. Um, and I'm gonna show some examples of where that can really cause a problem. Um, increases react uh reactive workload. Yeah, I mean if something you you're not in control of this. None of us are in control of uh if an AI um video or uh fake individual springs up, it's it's difficult. You've got to react and be able to react, and the ability to be prepared and react is is why um you might want to um have a clear plan of what you're gonna do. Um and obviously your plan's got to be quite flexible because all sorts of bits and pieces uh outside variables are outside of your control. It can dilute the your brand tone. So I'm gonna show you a couple of examples of where um client clients have done campaigns and they've actually become gone out of their control. So it's it's affected the brand. It can it raises reputational risk, and in in a modern era in 2026, reputation is worth more than anything. If you've got if you haven't got a reputation, you're ruined basically

Why AI Misinformation Hits Hard

Chris Norton 6:24

in terms of stock price or whatever it is. Um it undermines visual storytelling, especially if it's it can help storytelling as well, and AI, but um obviously with it with misinformation not so much, and it can damage credibility by association. So even though um you haven't done it and it's about you it's about a brand or a category that's similar to you, it can it can affect you um by association. So AI misinformation is now constant, it's happening all the time, like constantly. Um like hacking is it's cheap, um, really cheap. It's getting cheaper by the day, um, and it's convincing. Like if you are sat there thinking, oh yeah, I can tell an AI video, I'm gonna show a couple of examples where you can. You know, the McDonald's example I've got in a minute, you'll see, and you're like, oh yeah, I can see that's AI. But that's a year ago, uh, things are moving on. Um, brands are expected to respond instantly. So everything, everything's instant thanks to the Amazon effects. We want delivery tomorrow. We want delivery with it, yeah, we want delivery tomorrow, and now within eight hours, we expect everything to be done within 24 hours. Um, and silence is often interpreted as failure. So if you don't if you just let something happen, sometimes silence can be a good thing, but often in this sort of space, excuse me. I'm glad you found that amusing, Will. Uh, silence is interpreted as failure. Umly nine percent of 9% of consumers feel confident identifying a deep fake. So, yeah, less than 10% of people feel confident in identifying it identifying a deep fake. And in some of our workshops, I was showing the Tupac Shakur ones with Tupac and Michael Jackson rapping together. You know, obviously you know that they're fake, but a lot of the time you can't tell. So, and just to show you, this isn't a deep fake, but this is just to show you something where uh that I prepared. This took me about a minute to prepare, and this is if I wanted to call Will on the phone. Hi, Will, it's Chris. I have just had a call from a supplier, and they need us to pay for some event costs up front. I am out and about at the moment. If you don't mind, can just pay them online by bacs for me. Their account number is 789987, and sort code is 203862. The amount is 24,999. Thanks for helping me out, mate. I owe you one.

Speaker 4 8:44

Now Did I pay it? That's the question.

Chris Norton 8:48

You would as well, probably, although you're pretty tight, so I doubt it. Um, yeah, so the the the the point there is you can start you can tell, I would imagine. I mean, we've done I've done we did a podcast where me and Will used it had fake me and him, and we told people it wasn't us. Now, obviously, that was done from a 30-second clip of my voice, but if you put in like an hour's clip, it's gonna have more iterations and uh the right inflections at the right places and sound more emotional. That took me two minutes to do. So if someone can send you a what your, you know, your a member uh of your staff, uh a WhatsApp from a number that they don't recognise. I've lost my phone, sorry, can you send me this? You can see how easy it is to deep fake, and it that was free as well, as a free test. But the point is you can do that in five minutes. So um 27%, so more than a quarter of generative AI misuse is devoted to shaping public opinion. That's according to the Financial Times. Um, I I mean you could argue that there's a whole thing in America. America have got um a behavioral science team that that uh specialize in nudge, whether that's misinformation or whatever, there is a team dedicated, as and they apparently, when we interviewed someone on the podcast, they told me that um they copied the UK, which was quite interesting. UK was the first to have a nudge behavioral science team, uh, and it that it can be done for positives, you know, such as a great thing with getting vaccinated, etc., or negatives to get you to spend your money. So trust has shifted then. Customers now trust screenshots, peer-to-peer sharing. So think WhatsApp and think the message I've just shared you. Um visual evidence, you often you believe what you see, and but people are becoming more skeptical, which is a good thing. We need to be more skeptical. Um, influencers, people believe influencers more than ever, hence why we've got an influencer uh event later on in the year. They're still really effective because it's you know um uh content that's not generated by a brand. Um, private platforms drive belief faster and mainstream. Now, um I was actually at the weekend I was I went to watch Back to the Future um the musical, which is absolutely brilliant, by the way. Best thing I've ever seen on stage. I've seen loads and loads of different things, but brilliant. Um totally shocked because I thought it was gonna be rubbish. But we took my son. Anyway, the point is this is a little clip I've got from Back to the Future with a twist. So this is Back to the Future, but not Marty McFly is Tom Holland and Doc is Robert Downey Jr.

Speaker 7 11:23

Are you okay? Isn't he a greenball?

Speaker 6 11:37

This is more serious than I thought. Apparently,

Trust Shifts And Voice Cloning

Speaker 6 11:39

your mother is amorously infatuated with you instead of your father.

Chris Norton 11:43

So the point there is there's a whole clip on YouTube which we can send afterwards. Um the timing, I don't know, it was a little bit it's it's fine actually when you watch it online, um, but obviously with the there's a little bit of a lag, but the the timing's perfect. The the point is they've swapped the characters to be and and that can be done seamlessly, so you cannot tell. Uh completely different actors. It's crazy what it and it was all to show the risks for Hollywood and and deep fakes. It's uh it's a real problem. And the truth is, Hollywood brands were losing control a bit of the narrative and how to do that. So, why are brands losing control? Well, um, creation is instant, like it is so fast. To do um, there was a Tom Cruise video that I shared on one of our workshops recently, and that was from about four years ago, and it needed a high-end um AI specialist, it was a high-end um visual specialist, it took him ages, but it they used a Tom Cruise impersonator, and it was Tom Cruise, right? That was four years ago. Now you can do it instantly. Uh, and that's the problem, is is that it's just so instant, and you don't need an expert to do it because the technology's just got better and better, or worse, depends on how you see it. Um, paid media uh accelerates reach, so they can create a video and they can ex they can put some $100 behind it, and before you know it, it's got to 10,000 views, especially if it's entertaining and shocking, which a lot of AI videos can be. They're often um provocative in nature. Um, screenshots outlive takedowns. So if there's a problem, you want something taken down, you can object to it, it could take 24-48 hours to have it taken down. But often people just take a screenshot and share it, and then it's on the internet forever. Um private channels are invisible, so yeah, that we used to call that dark social, but you've got private channels, and it's often political or humorous, so you can't really control it. And humor, we know, uh, or political creates anger and outrage. Uh, other stuff that does the best, which is why they would use it. So, how does AI spread then? Well, it creates believable content that you you're generally watching, thinking, is that real? Is that not real? And you I know you'll be scratching your head thinking of things that you've seen recently, you're not sure if it's real or not. Especially some of the Iran stuff that's terrible, obviously, but some of that is is not real as well. Um, paid and organic amplification, screenshot capture, which we've covered, a bit of private sharing, and then your your brand reacts. So that's kind of how it how it occurs. It's sort of believable content, it gets amplified, somebody screenshots it, it gets shared multiple times on WhatsApp, and then right at the end, you it's been on, it's been going viral for five hours, you spot it, and someone's tagged you in it, and and the brand reacts, and by then the the damage could be done. So it's understanding the risk, the trends and emerging threats to make sure that we don't get caught, you know, um in a in a compromising situation. So, what are the key AI threats and trends then? Well, let's start from the top and we'll work round the clock there. So we've got fake influencers and AI personas. That's that's a thing already. We've got voice cloning and audio, deep fakes, which I just showed you. Um uh uh a voice clone of myself and to Will that took 30 seconds to make. Well, it took a 30-second clip of audio to make. And if you've got a video online of you talking, they can make a video of you, they can make a deep fake. Deep fake video ads, which we're gonna show some examples of in a minute. Uh, so celebrities being used in videos that aren't necessarily correct. Uh, search generated misinformation,

How Brands Lose Control Fast

Chris Norton 15:13

so articles that are created to create misinformation, a classic tool in used in propaganda and war and stuff like that. Fake reviews, people using fake reviews on Amazon or wherever, or hotels to dis discredit other restaurants or hospitality. That's a that's a problem in itself. Scam events that you sign up for, pay a registration fee, you go to attend, and guess what? It's not there. Um, brand impersonation. So people uh impersonating your brand setting up the same URL as yours with a hyphen or with a dot org or with a dot, you know, you can keep going. Um seen tons of those. Um and we've got we've got ways to beat that that um in our crisis um sessions, and then uh global regulator um pressure as well. So that's there's some of your key threats. So what are the challenges then? Well, um AI tools are widely accessible, they're and they're updated every day. It's a it's a space race. One one one um platform brings out a new tool, then the next day another platform matches it. Deep fake quality has surged. It's brilliant. I mean, I it's it's difficult. I've seen a few videos where I've been like, is that real? Is that not real? Um, I do think that these social channels need to get on this. I think they can use technology to improve it. Meta in particular could pull, you know, this they should all there should be some governance around this, but it's not there yet. Algorithms reward high engagement. And when we say algorithms, if you imagine a TikTok video, people get paid for views, they get paid for engagement. So if they get loads and loads of engagement, they're gonna plug loads of um videos that anything that's gonna get engagement, and if they can create it with an AI prompt, they're gonna do so. Uh or something controversial. People like to be controversial, you know, for any old reason, just to just to get engagement and then to sell products off the back of it. Consumers are overloaded with content, that's true. There's so much content, much more and it's more is flooding onto the internet, and regulations I've put hasn't, it should be haven't caught up, they just haven't caught up. Regulations are still miles behind, and they always will be with something like this, I think. But they could they need to get on top of it. Um, multi-market brands face um amplified exposure. So, yeah, if you're in multi-different uh several different territories around the world, like some of our clients are, it's a real risk

The Main AI Threat Landscape

Chris Norton 17:32

because something can pop up in a in a territory that's not as well protected, and and we've seen that problem already. So, according to Experian, 64% of businesses reported experiencing rising fraud losses, including this is including AI attacks. So, riding fraud losses include that's like hacking and AI attacks in 2026 already. And 35% of businesses were targeted by AI-related fraud, so a third um in Q1 last year. That's up from 23% in the previous year, and you can be guaranteed at the end of this quarter it will be up again this year because it's easier and cheaper, cheaper to do. So let's look at when the internet has um used its humour to hijack a brand campaign. We all love a marketing campaign. Uh, here's a campaign from um uh uh Walkers that used Gary Lineker as the obviously the he's the uh spokesperson, and they used him uh to create videos, and people could then upload their picture and it would be a bespoke video campaign. Um, but it the internet just got involved, and this is what happened.

Speaker 6 18:40

Thanks for joining the Walkers Wave and celebrating the UEFA Champions League final. Nice selfie. Here you are, uniting with other fans outside the National Stadium of Wales. The Walkers Wave, a thing of beauty.

Chris Norton 19:04

As you can imagine, there's another one just before that, but I'll I'll let this bit play. Um, yeah, so basically there was uh it was Rolf Harris, it was uh Jimmy Savile, uh, tons of them. And it basically uh the internet was just having a bit of fun with it, and unfortunately, Walker's lost control of the campaign. Then you've got Jet2, the Jet2 meme. For those of you, I mean, uh, Jet2 has gone everywhere with that um with the quote from the lady nothing beats a Jet2holiday. So then people started sharing the worst situations possible on holidays for Jet2.

Speaker 2 19:37

Nothing beats a Jet2 holiday, and right now you can say £50 per person. That's £200 off. And then the next one Nothing beats a Jet2 holiday. And right now, you can say £50 per person. Do you make me laugh, guys? That's £200 off for a family of four.

Chris Norton 19:57

He looks in genuine d distress. So the point is you can't control your campaigns as much as you wanted to anyway, but um with AI you can do even more. Um so there's organizational challenges. Um, there's a private risk assessment for digital. So you private risk assessment for digital campaigns is crucial. So when you're doing a digital campaign, maybe thinking, could this go awry? What could people upload and cause trouble? But even the most positive campaigns can be hijacked and changed. Um importance of real-time tracking and monitoring. Yeah, you've got to stay on top of it to decide whether you want to continue. So when we I said before that we had deepfaked audio, we've got deepfake videos as well. So those of you that you that will know this guy, this is uh MrBeast, he's the biggest YouTuber on the planet, and he he makes a lot of revenue uh in the millions for his videos. Um, but an AI deepfake video of him appeared saying that you could get um an iPhone for two dollars, which just seemed ridiculously cheap, directing viewers to phishing sites, so then they put their details in, they'd capture the data, and then lo and behold, you'd get you know you'd get ripped off. Here's the video and see what you think.

Speaker 21:02

If you're watching this video, you're one of the 10,000 lucky people who'll get an iPhone 15 Pro for just $2. I'm MrBeast, and I'm

When The Internet Hijacks Campaigns

Speaker 21:10

doing the world's largest iPhone 15 giveaway. Click the link below to claim yours now.

Chris Norton 21:17

I mean, that's pretty real. There's no and it was caused a big problem because a lot of people got scammed by that. So, yeah, a bit of a one to worry about. Um, so you've just got to be careful in what where you put your data. It's just people just got to be more vigilant than ever. But for us as brand owners, we just need to be thinking about how our campaigns can be used against us if they if indeed they will be. Then we've got brand created AI. So this is an area that you guys might have your own views on. Feel free to share in the chat because when Will's section comes on, I'd love to hear what you think about brands using AI to create amazing content. So in December 2025, um McDonald's Netherlands posted a fully AI generated Christmas ad that is Viewed as creepy, soulless, and lacking authenticity. They have depicted holiday mishaps and used AI visuals. So let's just see if I'll show you the video in a minute. But even when the brand created AI content is a legitimate poor execution, the risk isn't only external misinformation, it can also be from generated content by AI by a brand. And if it looks a bit naff or the ideas bad, let's see what you think of this. You get the gist. So um interesting. I don't think it's I mean, obviously it's a more recent example that, and you can't really overly tell that it's all AI. Some bits, obviously, you know, you could say a CGI, but yeah, interesting. It but it just didn't feel very authentic. And also, I'm not sure the idea is a brilliant one. Most terrible time of the year. I know they're trying to be controversial, but it was interesting. Yeah,

Scam Ads And Brand Made AI

Chris Norton 23:26

that was pretty derided that that campaign. But uh sorry if anyone was involved in it. Um, and then we've got here like H&M announced plans for an AI-generated digital twin models. So you, in other words, have you m the models having a digital twin? Um, so using AI models to do modeling, and then there's the all the arguments over ethics, um, replacing jobs or models because it's cheaper. Let's see what you think.

Speaker 3 23:51

H&M is creating digital replicas of their models. The fashion giant plans to develop AI twins of 30 models this year, offering them full ownership of their virtual counterparts. These digital avatars can take modeling jobs while the human models are elsewhere, potentially doubling their income opportunities without additional time commitments. When one model first saw her AI twin, she described it as exciting but unsettling, a sentiment echoed across the industry as fashion grapples with AI. H&M acknowledges its approach will be controversial, but believes involving models directly in the process protects their rights and livelihoods as AI inevitably transforms the industry. The company plans to watermark all AI-generated content for transparency and to gauge public reaction. Critics worry about impacts on photographers, makeup artists, and stylists whose jobs could diminish if AI replaces traditional photo shoots. By merging human talent with digital innovation, H&M aims to establish ethical standards for AI and fashion while the industry navigates uncharted territory.

Chris Norton 24:46

I mean, I'm not sure what I think about that with copyright and fake people. What do you think, Will? Do you think that's a good use of like because of the copyright infringement and like the fact that people's images are being used? What do you think?

Speaker 4 24:59

Well, it's in some respects it's progressive because they're involving the models and they're also fully disclosing the use of AI, but it is a strange uncharted territory, isn't it? The fact that actually one model could be doing five or six jobs at the same time. And actually, I think no matter how good AI is, it of you know, like that McDonald's ad, it it it there's something about it that doesn't feel quite right, and it just doesn't, you know, it's the um what's the phrase, the uncanny valley effect, isn't it? It just doesn't ring quite true. So progressive on one hand, but uh yeah, strange this is the worst AI is the worst AI is ever gonna be.

Chris Norton 25:39

That's the issue I've got.

Speaker 4 25:40

Um day, we maybe one day we can deliver out um webinars as AI avatars, Chris. We probably could today if we tried. While we do the day job in the background.

Chris Norton 25:50

Um deep fake corporate c fraud then. So Arup um was defrauded of £20 million after scammers used a deep fake video call featuring realistic voices, images, uh, and misleading uh an employee into authorizing transfers, similar to what I was doing with Will, but on a much bigger scale.

Speaker 1 26:07

So if we just click this, um just last month, a finance worker in Hong Kong got scammed into paying a whopping $25 million to scammers who used deep fake technology to successfully pose as a company executive during an entire Zoom meeting. Now, Hong Kong police have yet to reveal the identity of the company or the worker who got scammed, but here's what we do know based on the reporting. The worker initially received an email from what claimed to be a UK-based company chief financial officer, the CFO. But the finance worker was skeptical because the email was super fitchy. Red flags were raised after the email referenced a secret transaction that had to be carried out. But then things got even weirder. Okay? The worker was later persuaded that no, no, no, this actually isn't a scam at all. Because during a Zoom call with the CFO and other colleagues they knew, the worker ended up getting persuaded that, like, no, I'm talking to a real person. This is this is legit, right? And so let's move forward with this deal. Now, um, the worker was later persuaded. However, let's talk a little bit about what happened here. So here's what the Hong Kong police uh suspects happened. Scammers found publicly available video and audio of the impersonation targets via YouTube, and then used deep fake technology to emulate their voices to lure the victim to follow their instructions. But that's only one piece of the puzzle, so there's more. The fraud in Hong Kong almost certainly used real-time deepfakes, meaning that the fake executive mirrored the scammer as they listened, talked, and nodded during the meeting. According to David Maimon, a criminology professor at Georgia State University, online fraudsters have been using real-time deepfakes on video calls since at least last year for smaller scale fraud, including romance scam.

Chris Norton 28:17

So, in other words, they used instant like now. I c I could become that person and talk, and it would be that other person on the call. Instant real-time deepfake to trick. So the person could just act and then it looks like the person that they want to speak to. Amazing but scary technology. Uh so it shows that deepfakes aren't just misinformation, they can directly cause financial loss, um, uh operational dis uh disruption to other organizations, and brand and reputation affects bleeding to customer trust

Deepfake Fraud With Real Money Loss

Chris Norton 28:48

and internal security. So a quite a worrying scenario, but you can prepare for it. But first, Will's gonna go on to um how to prepare for it. But before um Will's gonna go through the polls.

Speaker 4 29:01

Yeah, so fascinating to see your responses to some of these points. Um, so I'm gonna share them on the screen and then we'll have a quick chat about them. So is your crisis plan AI and social media ready? So um typically people are starting to incorporate social media into a crisis plan, which is great because you know, a year ago people weren't even taking into consideration social media as part of a crisis plan. But I think AI needs to be a consideration there. So either it doesn't consider them or it's social only. So that's quite interesting. Um, who owns misinformation in your organization? Uh well, this actually plays into what I'm about to talk to. So marketing and comms, along with everything else in an organization, typically things get delegated to marketing and comms, don't they? But actually, AI as AI and misinformation does sit quite logically with marketing and comms. Um so uh your what you know what what best uh describes your approach to misinformation, so neither proactive nor reactive, which implies that people aren't yet thinking about it, which is fine. That's why you're on this webinar. Um, and people tend to be reactive when you know when it happens, you deal with it, but you're not proactively taking steps to deal with it. Again, that's fine. We're gonna talk through how you can be more proactive. And it is an emerging discipline, AI-driven misinformation. So it's you know, it's it's it's you know, don't beat yourselves up about it if if you're not doing anything yet. Um, what are people most worried about when it comes to misinformation? Um, search-generated misinformation, brand impersonation and parody, fake reviews. Yeah, all of these are you know, we we we have conversations with clients about these challenges frequently. And then have you suffered um from some kind of misinformation issue or threat? So, yes, malicious slash coordinated, yes, but not necessarily malicious. And that's that's that's the reality of a lot of misinformation. It's not all, you know, that jet two. I know that was a bit of a parody thing, you know, that that wasn't necessarily driven by hackers or anything. It was it's just a bit of fun and something that gains traction. And often this stuff can can be not necessarily malicious, but can equally cause a lot of damage. Often it is malicious, and like you know, like the MrBeast example, it's about data capture or it's about um you know uh hacking into people's bank accounts or whatever. But yeah, certainly interesting. So um we've um alarmed you with uh with the um with uh the the problem and how much it's gonna grow. Now we're gonna talk through um some of the solutions. If I can stop sharing this. So this is how to be proactive and how to fight back. Um and essentially it's a framework. So it we what we what we're gonna look at now is how you can actually adapt your crisis management plan to to take into consideration AI and misinformation.

Poll Results And Ownership Questions

Speaker 4 31:55

And essentially, this is you know, I keep talking about this, but this is about being proactive, not reactive. And it, you know, if you're interested in these topics, as Chris said, we um we have a uh a podcast that's um really well listened to. It's the number one um marketing podcast in the UK, a little plug there. But we we've we've tackled misinformation in various guises um over the last year, and there's a really great episode with Ant Cousins. I don't know if you know Ant, but he is the VP of Product at Meltwater, which is a kind of AI-driven social media listening tool. Um, and he specializes in AI and comms. And it's a really interesting chat. He he he really puts a lot of kind of meat on the bones and highlights how this is a growing threat. And and one of the key takeaways, and we're really on the same page with this, is you know, you as a big brand or as a brand, just being reactive is no longer an option. Um, you need to be very much proactive when it comes to the threat posed by misinformation. And it's almost a it's almost an insurance policy in the way that you you no doubt you've got a robust crisis plan, you know, we need to also have a robust misinformation plan as well. We're gonna talk through over the next few slides ways you can do that. So I found it quite interesting that 50% of you said um misinformation sits with marketing, and and from our experience, that that's that's absolutely right. Um, if you look at an organization, I mean I'm not saying these parties aren't involved at all, but IT they they see the world through systems, legal sees the world through liability, but from a marketing perspective, you know, marketing's uniquely placed in as far as you're aware of customers and customer sentiment and content, and you're aware and are able to treat and for that reason, I think marketing tends to own misinformation, and therefore the board, the C-suite, will look to marketing to understand how you're gonna mitigate against misinformation. So, um, yeah, um, we need to be prepared and we need to build our knowledge. So it it's you know, as ever, with a lot of these things, it's not necessarily rocket science, it's more about improving current processes. So I'd love to know in the chat actually how many you how many of you have got a kind of a robust crisis plan and how often you use it and refresh it. We're we're big advocates in regular scenario testing when it comes to a crisis plan. You know, just having a crisis plan isn't enough. You need to be constantly testing it and stress testing it and improving it. But essentially, um, you know, the crisis process, you you've got your pre-crisis phase, which arguably is the most important phase. This is all about preparation, it's all about um having a robust

A Proactive Framework That Repeats

Speaker 4 34:37

plan, constantly updating it, scenario testing it, having the right the right protocols in place. The better your pre-crisis plan, the the more effective the next stage is going to be, which is the crisis response. And ideally, you won't you won't be called into action too often. But when you do get a crisis, essentially you've then got to implement the crisis plan and you've got to try and mitigate that crisis as quickly as possible. And then another really important phase of crisis management, and a lot of people forget this, is the post-crisis phase, which is where you critically analyze what worked, what didn't work, and all of those learnings can go back into that pre-crisis phase, and it becomes this kind of loop of continual improvement. So we're going to dig into some of these stages and look at the role that misinformation and AI increasingly plays in each of these. So crisis response. Um crisis response. So two things to bear in mind. First of all, we need a repeatable framework, um, i.e., we need a robust crisis plan that is um can be used in the same way um multiple times. Essentially, the first 24 hours in terms of how you respond to a crisis um are super important. So um speed tends to beat perfection. Um by that I mean don't procrastinate endlessly over what your response is going to be, because what that will do is create silence and it will create um a vacuum where you get suspicion and speculation. So any kind of response is better than no response, and this kind of calm clarity um tends to protect trust. You know, don't sound try not to sound defensive, try not to sound panicked. That kind of goes without saying. But essentially, what what you're looking at with this kind of um crisis plan is is a repeatable framework. So um it allows you to do things like um you know pause marketing activity, which I'm going to come on to. Um you know, have you got kind of a campaign running that is is at odds with with what's happening? You might issue a holding statement and you need to also monitor and update, um, you know, monitor what's going on and update the team. When you issue a holding statement, you need to be really clear. So what you know, what is false information? You need to reassure your customers, you need to update and avoid kind of legal language and corporate cliches. But essentially, you know, we talk about the golden hour or the golden, you know, the golden 24 hours. It is that first kind of 24 hours that's really, really important to get right when it comes to a crisis. Um, I've realized this is slightly out of order, so apologies for that. Um, I'm gonna jump back to pre-crisis crisis now, which is all about the preparation part. So the more and and just to reiterate, the more time you spend in pre-crisis, the more effective your um your response when it actually comes to uh uh delivering your response, your your crisis response is gonna be. So pre-crisis, we need a crisis plan. Most of you will have a crisis plan, but we need to make sure it's social media um enabled, which it sounds like by the polls, most of yours are, but also we need to look at potential AI scenarios that might occur as part of this crisis plan. So Chris talked about things like fake influences, scam um scam promotions, see uh you know, CEO um mimicking or or deep fake voices. All of these scenarios could affect your um your brand. In fact, the the training um session we most recently did regarding um misinformation, it was all about um fake um fake nutritional information on some of the products, and there's lots of misinformation from

The First 24 Hours Response

Speaker 4 38:24

activists and from um people wanting to to uh damage the brand, um, spreading rumors that uh carcinogenic um ingredients were included in their products, which just wasn't true. What's really important as well is things like defining governance and ownership. Everybody needs to have very, very clear roles, as with any um crisis plan. You need to be really clear on who does what. This idea of testing response simulations is really important as well. So when we've agreed the different scenarios that might occur, we need to then look at how that might play out and and you know look at who who who who deals with what, how statements are put out, um, you know, what happens with um with social media, what happens with your website. Um, and we also need to monitor visuals, not just text. And this comes back to the idea of being proactive. So when we spoke with Ant Cousins, his tool at Meltwater is a social listening tool, but it's it also monitors visuals. So it's not just about looking at how your brand's talked about, it's actually looking at how your products are being shared on image sharing sites or how your logo might be being used. And a lot of misinformation um is visual in nature, not text in nature. So you really need to make sure your your kind of social listening is up to scratch. And this is something I'm going to talk about later, but the way we work with influencers needs to be tightened up as well, and we need to bear in mind that influencers may inadvertently be using AI content, um, or we might get fake influencers or whatever, as the MrBeast example showed. And then we need to anticipate risks, and you know, this this comes down to being a little bit creative. You can actually, ironically, you can actually use AI to predict some of the AI-generated misinformation risks you're likely to find. So you're going to have two types of risk, always on risks,

Pre Crisis Planning And Monitoring

Speaker 4 40:15

which are things like CEO impersonation, scam promotions, fake news articles, but then you might have campaign-specific risks. So you might be launching a new marketing campaign, a new product launch campaign, and um there may be specific nuanced risks related to that campaign. Back to the Jet2 example, it it's in a lot of cases, it's very difficult to anticipate the risk. I think when Jet2 launched that campaign, um, with the best will in the world, you'd never guess that people are going to hijack it and start talking about holiday nightmares. So you can't always do this, but as far as possible, try and anticipate um different risks. And it might might not be a case of cancelling the campaign if you do think there's a risk, it's more about being prepared should that risk happen. It's worth here looking back to the kind of um, you know, the core principles of online crisis management. These are still absolutely applicable. Um, you know, and and and AI doesn't, AI and misinformation doesn't fundamentally change the way you approach crisis management, but um, you know, certainly um you need to make sure um you are uh in line with best practice. So response, obviously, act quickly, focus on facts, use your own channels as well, where there's less risk for um for for you losing control of the message, so that um i.e. your website or or whatever that may be. Messaging, of course, you're gonna have pre-approved statements, and those pre-approved statements are going to take into consideration misinformation as well. Communication, you need to make sure your um your internal comms is up to scratch. You also need to make sure your internal teams are aware of AI misinformation, they know how to spot a fake, they know how the social listening tools work, they know they you know they know what they're doing in the context of misinformation, so nothing is a surprise to them. And then consistency as ever. So regular updates, monitoring sentiment, etc. etc. So this remains really, really important. Um, we're gonna talk about influencers in a second, which are really important, but I guess um by way of conclusion, crisis quick tips, and and again, back to what Ant said, we need to be proactive. So identify issues early, and you can do that through um through training, through skills and development, through using the right tools. Proactive social listening, taking into consideration audio, video, visuals, as well as text only. Be ready to secure your social media accounts if you need to, of course, pause um, you know, pause scheduled posts or campaigns, be great at internal comms and make sure everybody knows everybody's on the same page. Know when to engage and when not to. And that's one of the kind of the golden rules of dealing with a crisis in social media. If you're getting 10,000 complaints about something, it's not possible to necessarily respond to all 10,000, pick out the top themes and respond and pin those to the top of the profile, whatever it may be, um, and go public quickly if required. Again, silence equals speculation, and it creates a vacuum. So even if the statement is we're looking into it, issue a statement. So influences and misinformation, um, and these do pose a unique set of risks, and and and this really needs to be a consideration. The way we work with influencers and the way we uh we risk assess working with influencers. And um I suppose the the the risk of influencers is particularly if you're working with kind of bigger ticket influencers, um they are by their nature in general trustworthy. Consumer, you know, all sorts of research shows that consumers trust the views of influencers above traditional media channels. Channels, they have the potential for messaging or content to spread extremely fast via the social media algorithms. They can often kind of blend into authentic conversation. It's not always immediately obvious if they're an influencer. And the biggest challenge is that influencers are firmly on the agenda and on the radar of the mainstream press. So if an influencer does something untoward or we face a fake influencer, the chances are it's going to be newsworthy. And what you get is this kind of cycle where you know you might work with an influencer and something might happen that's bad or some misinformation might get spread. They've obviously got a huge community around them typically. The media are one of those, and the bigger their audience, the more newsworthy they're deemed. You might then get a controversy or some sort of complaint about their content, and that in turn might create regulatory risk. And you can see how it can quite quickly spiral. You know, as soon as the mainstream media who are already aware of these influences get hold of it, you're in trouble. So there are ways you can kind of mitigate against this. Now, what we're talking about when it comes to the types of risks, it can be anything really, but the the types of things we're seeing the most of cloned influencer accounts. So typically an influencer will get um, you know, somebody will set up and grow a fake influencer account. So in some cases, that could mean that influencer who's who's a bot may get in touch with you, you might gift them product, you might even form a brand partnership with them and it's not even them. Or a well-known influencer might start talking about an issue or a topic related to your brand that simply isn't true. Equally, there's an emergence of fake talent agents who um who essentially are on the scam, on the take, and they will work with um apparently work with influencers, but it's not really them. Deep fake ads promoting scams, again, that's down to the MrBeast example. Typically, that comes down to illegal data capture. And also, you know, see CEO impersonations and things like that. I don't know if anyone saw the McDonald's um CEO eating the burger, which we all watched on Monday as part of um as part of a team meeting. That was that was real, I believe, but yeah, you know, that quite easily he actually looked like an AI created bot, didn't he, the way he was eating the burger. So

Influencers Contracts And Due Diligence

Speaker 4 46:24

um, but yeah, there's a lot of risk out there um from influencers, and and the issue is even if you're not directly targeted, um consumers may associate the scam with your brand. So if you've got a fake influencer talking about um the carcinogenic uh properties of your burgers or whatever it might be, even if it's not true, it's gonna create you issues and it's gonna create you damage. So, what can we do? Uh if anyone works with influencers, no doubt you will do a due diligence process before you actually work with them, and that will involve probably chatting with them, looking at how engaged their community is, looking at their content. We just need to kind of expand that when we do our due diligence to make sure. Um, I suppose um, you know, one of the main risks of working with a legitimate influencer is um do they ever use um AI as part of their content? If they do use AI, um do they disclose it? You know, what's their have they got an ethical approach to using AI? You know, the the worst thing would be for an influencer to trial your product, but it's it's them using the products but created by AI, which immediately undermines them and it undermines your brand. We need to make sure as part of influencer contracts, we've got elements of contractual safeguards. So, you know, uh working with an influencer, typically we would lay out a really detailed SLA in terms of what we want from them. So, as part of that, we now say we don't want any undisclosed AI-generated content in the context of our product or our brand. Pre-approval clauses, claims verifications, that kind of thing need to be taken into consideration. And as ever with influencers, risk score the influencer based on what you find out about them. So if they do use lots of AI content but they've promised not to, you might be okay. But risk score them and make decisions based on that. But um, as with your crisis plan, when you're vetting influencers, bear in mind AI and misinformation risks. Um now, again, we need to be proactive when it comes to how influencers might be talking about our product or might be sharing misinformation. Again, social listening is really important. Um, we can um a lot of social listening tools now are AI powered, so they can identify AI content, so they can use AI tools to kind of supercharge your listening and essentially gives you much greater visibility of how your brand or um your category is being talked about. Things like keyword and image alerts can can can help reverse image or video search again, product and use logos, um etc. You can look at how they're being talked about. Um and essentially, over time, it's worth building a bit of a dashboard where you can not, you know, and this can this can be a way to kind of expand your existing social listening dashboard if you already do that, and also monitoring cloned accounts. If you identify problematic accounts that are uh brand parodies or whatever it may be, monitor them and keep a close eye on them. So post-crisis and long-term resilience. Um, this is the the kind of concluding section. I mentioned again, um, you know, what you do after a crisis is every bit as important as how well you prepare for it. And the more you can kind of analyze how an issue or or a crisis uh was handled, that allows you to constantly improve. And this is all about kind of continual improvement. So, you know, post-crisis or post-issue, how effective was your response? You know, which channels performed best, you know, which were the most positive, neutral, or negative channels, which channels were you able to kind of shut the issue down on quickest, reach an impact of the issue. That's your key metric for how how big or how small something was. Um, again, you can use social listening to track sentiment of an issue over time. So, how long was it negative for? How long did it take for your brand sentiment to return to normal? And start to update any, you know, any gaps where you lack governance or you lack clarity. And this is really important because um the more you do this, the the the better you will get, essentially, and and ultimately you're gonna get better, you know, the more you analyze it, your your responses are gonna be much, much more effective. And the more you incorporate misinformation and AI and and whatever else is around the corner, the more um effective your crisis plan is gonna be. So um AVU as an agency, and and this isn't just kind of predictions and this isn't just anecdotal feedback, but the the volume of conversations we're having with with global brands about AI information, uh AI misinformation is is growing. It's very much moved perhaps a year or two ago from something that was quite hypothetical to a very real threat. It's not always going to be malicious. Sometimes it may be um somebody creating content just for fun that spirals out of control. But the reality is it's it's a risk, and it's only going to become a bigger risk as AI technology grows. It's operational, reputational, financial, and cultural. And essentially the brands that don't take this into consideration um are going to be the ones most at you know most at risk. So be proactive, look at your crisis plan, take it into consideration, educate yourselves and educate your teams. So we do have a product. This is the the sales bit at the end. But um, if you're interested in some of these topics, we we can help you. You know, we can help you get AI ready, essentially. So no doubt you'll have a crisis plan, but a lot of the work we're doing at the moment is is in um enhancing and developing crisis plans with AI misinformation uh clauses, and essentially it's it's it's an insurance policy, if you like, against um the threat posed

Post Crisis Learning And Resilience

Speaker 4 52:30

by AI and misinformation. Obviously, it's bespoke to your business and sector, and essentially it's an updated crisis plan with a really detailed AI risk assessment. So, what are the risks your particular brand are likely to face? What are some example scenarios? We would then enhance your response protocols, we would develop a live monitoring toolkit, and that's bedded in via um scenario testing, which essentially is posing these three or four credible misinformation AI scenarios that you're likely to face, and it's putting your crisis plan um under pressure. So if you want to be ready uh for the risks posed by misinformation, do give us a shout and we can tell you more about it. And hopefully we haven't scared you too much, but um we wanted to end with a positive. But thank you very much for that. Hope you found that interesting. Um, there's been some great conversations in the chat. Thank you for filling in the polls, that's been insightful. Um, we'll send this through, as Chris said. We do have a whole programme of events. The next one's on influencers. So if you want to look at our um our on-demand events, we've covered all sorts of topics in the past or register for our future events. Have a look um there. There's our contact details, and also um, I would urge you to look at our podcast as well. Um, as I said, we've covered AI, we've covered misinformation and all sorts of other topics in in all sorts of detail. So uh lots of um interesting content.

Chris Norton 53:55

Number one marketing podcast in the UK, William, as of December.

Speaker 4 53:59

Very good. What number are we? Top is it top number three business podcast in the UK?

Chris Norton 54:04

We were just next to Stephen Bartlett the other day, number two, but we're number four at the moment. So if you look at that there, you can see we're number four embracing marketing mistakes. I literally literally the first thing. I didn't I didn't rehearse that. Um, yeah.

Speaker 4 54:18

So yeah, thank you very much for joining us today. Uh hope you found that useful. Um I can see the chat's going mad, which is what we like, and yeah, you'll get copies of that. And um, any questions in the meantime, get in touch.