AI misinformation now costs brands real money and trust. Deepfakes, hijacked campaigns and cloned voices are everyday risks in 2026. Your crisis plan needs an AI upgrade.
A few weeks ago I sat down to record a webinar with Will, my co-director at Prohibition, on protecting your brand in the age of AI misinformation. To make a point, I cloned my own voice and left Will a fake voicemail asking him to pay a supplier nearly £25,000 by bank transfer. It took me about two minutes and cost nothing.
Would he have paid it? Probably not, he’s far too tight. But that’s the bit that should worry every marketer reading this. If I can fake myself in two minutes from a 30-second clip, someone can fake your CEO, your finance director or your biggest influencer just as fast.
This is the stuff Will and I now spend a chunk of every week talking to clients about. So here’s what I’d want you to take away from it.
What is AI misinformation and why does it matter to brands in 2026?
AI misinformation is any false or misleading content, made or spread using AI, that damages your brand. In 2026 it’s constant, cheap and convincing, which is exactly why it matters.
It erodes customer trust, it confuses people about your products, and it creates a pile of reactive work you never asked for. Reputation is worth more than almost anything a brand owns, and this stuff goes straight at it.
The pace is the problem. Social media made brand crises fast. AI has made them instant. We’ve reached the point where two updates to a major AI model can land within a day of each other, and the tools keep getting cheaper and better. Research from Google DeepMind, which mapped nearly 200 real cases of AI misuse, found that influencing public opinion was the single most common reason people misuse generative AI. This is not hypothetical any more.
How much damage can a deepfake actually do?
Enough to move millions out of your bank account. The engineering firm Arup lost around 25 million dollars after a finance worker was tricked by deepfaked colleagues on a video call.
The Arup case still stops people in their tracks when I show it. A member of staff in the Hong Kong office got an email about a confidential transaction and smelled a rat. Then they joined a video call where the CFO and several familiar colleagues were all present, talking and nodding. Every single one of them was a deepfake. The worker sent roughly 200 million Hong Kong dollars, about 25 million dollars, across 15 transactions.
It isn’t only the big corporate frauds either. A deepfake ad of MrBeast offering iPhones for two dollars ran on TikTok in 2023 and pulled people into a phishing site that harvested their card details. Fewer than one in ten people feel confident they can spot a deepfake, and that gap is where the money disappears.
Why do brand campaigns lose control so quickly?
Because the internet will always find the gap you left open. The moment a campaign invites people to add their own content, you’ve handed over some of the control, and AI has made that far easier to exploit.
My favourite cautionary tale is still Walkers. Back in 2017 they ran the Walkers Wave campaign with Gary Lineker, asking fans to send selfies for a chance to win Champions League final tickets. Nobody was checking the images properly, so people uploaded faces of Jimmy Savile, Rolf Harris, Fred West and Harold Shipman, and the system happily put them next to Lineker. Walkers pulled it and apologised. It kind of reminded me of Boaty McBoatface which was also rather amusing from us all on the Web.
Then there’s the Jet2 holiday meme, where that cheery “nothing beats a Jet2 holiday” voiceover got stitched onto the most miserable holiday footage people could find. No hackers, no malice, just the internet having fun at a brand’s expense.
Brands doing it to themselves is the newer wrinkle. McDonald’s Netherlands put out a fully AI generated Christmas ad in December 2025 called “It’s the Most Terrible Time of the Year”, and it got slated as soulless AI slop before being pulled. H&M went the other way and built AI digital twins of 30 of its models, with the models keeping ownership and getting paid per use. Progressive in some ways, deeply strange in others. Even when it’s well meant, there’s an uncanny valley to a lot of this that audiences clock instantly.
How do you build AI into your crisis plan?
Treat AI misinformation as another scenario your existing crisis plan has to cover. Most of the work is preparation, not heroics in the moment.
Will runs our crisis work, and his framework is simple: pre-crisis, crisis response, post-crisis, then loop the learnings back round. The pre-crisis phase is the one people skip and the one that matters most. That’s where you map the AI scenarios that could hit you, fake influencers, scam promotions, a cloned CEO, define who owns what, and stress test it before anything goes wrong.
When something does land, the first 24 hours decide how it goes. As Will put it on the day:
“Speed tends to beat perfection. Don’t procrastinate endlessly over what your response is going to be, because that will create silence, and a vacuum where you get suspicion and speculation.”
He’s right. Any clear, calm response beats silence, because silence reads as guilt. And your social listening needs to watch images and video, not just text, because a lot of this never shows up in a keyword search. When we had Antony Cousins, VP of Product at Meltwater, on the podcast, that was his big point too:
“Just being reactive is no longer an option. You need to be very much proactive when it comes to the threat posed by misinformation, and it’s almost an insurance policy.”
How should you handle influencers and AI misinformation?
Tighten your due diligence and your contracts. Influencers carry huge trust, which makes a fake or careless one a genuine reputational risk.
People trust influencers more than traditional media, and the press watch the big ones closely, so anything that goes wrong with an influencer tends to become a story fast. We’re seeing cloned influencer accounts, fake talent agents, and deepfake ads built off real creators’ faces.
So we now write AI clauses into influencer contracts. No undisclosed AI generated content involving our client’s brand, pre-approval on claims, and a risk score for every influencer based on how they actually use AI. The worst outcome is an influencer “testing” your product in content that turns out to be AI generated. That undermines them and it undermines you.
How to be proactive about AI misinformation
- Identify issues early through training, skills and the right monitoring tools.
- Run social listening that covers audio, video and visuals, not just text mentions.
- Be ready to secure your social accounts and pause scheduled posts or campaigns the moment something breaks.
- Know when to engage and when to stay quiet, and pin your response to the top of the profile.
- Go public quickly, even if all you can say is “we’re looking into it”.
- Add AI disclosure clauses to influencer contracts and risk-score everyone you work with.
Listen to the full episode
Watch on YouTube or find every episode at embracingmarketingmistakes.co.uk.
Embracing Marketing Mistakes is the weekly podcast I host with my good friend Will Ockenden, where we dig into the marketing and PR mistakes brands make so you don’t have to repeat them. This episode came from a live event we ran on protecting your brand in the age of AI misinformation.
Remember, this is the worst AI is ever going to be. So the real question isn’t whether a fake version of your brand turns up. It’s whether you’ll be ready when it does.
Chris Norton is the founder of Prohibition and an award winning communications consultant with more than twenty years’ experience. He was a lecturer at Leeds Beckett University and has had a varied PR career having worked both in-house and in a number of large consultancies. He is an Integrated PR and social media blogger and writes on a wide variety of blogs across a huge amount of topics from digital marketing, social media marketing right through to technology and crisis management.
