The media organization, TIME, working with ad-tech firm Mobian, has been testing “Agent Ads” — sponsored content written not for you, but for the AI crawler reading over your shoulder. Ally Bank, the U.S. digital bank, and the Project Management Institute, the global professional association, are already buying in. The pitch: skip the tedious business of earning credibility and just buy your way into the robot’s reading material. One problem: It doesn’t seem to work. And Perplexity has already called it “cloaking” and blocked it.

What the Hell Is an Agent Ad?

A normal ad is made for a human to see. An Agent Ad is made for an AI crawler to discover. It is brand-fed text on simplified pages that human visitors to the same site never see. Basically, they are planting information somewhere an AI might stumble onto it while researching, hoping that it shapes what the AI eventually tells someone.

It’s the old search-engine trick of showing a crawler something different from what a human sees. Which is why Perplexity calls it cloaking and has warned publishers using it they could get downranked. OpenAI and Google haven’t said they’ll treat it the same way, so it’s not dead. But nobody’s proven buying an Agent Ad buys a recommendation, either.

Of Course Marketers Want In

The stakes are real. Adobe found AI-referred traffic to US retail sites is up 393% year over year, converting 42% better than regular traffic. Similarweb found ChatGPT referrals to brand homepages jumped 354.7% in a single week after a UI change. Naturally, marketers’ first instinct after “how do we get in the answer?” is “how do we game it?”

The Robot Smelled a Rat

We’ve seen this movie. SEO began as making useful information easier to find before it curdled into keyword stuffing, link farms and cloaking, until Google got sophisticated enough to catch it, reportedly nuking 40 billion spammy pages a day. Agent Ads risk replaying that whole arc, faster, against a smarter referee.

Sorry. You Still Have to Do the Work.

This is what our PR for Robots sessions are about: making brands visible, understandable and recommendable when customers ask AI what to buy or whom to trust. We earn credible third-party coverage, develop real expertise and use SEO/AEO to make that evidence discoverable and legible to machines. Paid media can amplify a credible foundation.  But it can’t manufacture one.

The data backs this up hard. Muck Rack analyzed 25 million links cited by ChatGPT, Gemini, and Claude: 84% came from earned media. Paid and advertorial content? 0.3%. Eighty-four percent earned. Point-three percent paid. 

You Can’t Optimize Your Way Out of Being Uninteresting

Ahrefs studied 75,000 brands to find what correlates with AI visibility. Content volume barely mattered. The strongest driver was YouTube mentions, followed by brand mentions across the wider web — places you don’t control. AI doesn’t need more stuff from you. It needs credible evidence about you.

You Don’t Get to Pick Your Robot

There’s no single AI to please. ChatGPT’s share of AI traffic slid from ~76% to ~53% in under a year as Gemini surged. The engines don’t even read the same things. ChatGPT favors journalism, Gemini leans more on Reddit and YouTube, Claude seeks more academic output. Semrush found only 36 brands hold top-100 visibility across all platforms every month — Apple, Amazon, Patagonia-types with real, independent footprints, not just a nice website. Patagonia’s strength comes from being discussed on OutdoorGearLab, REI, GearJunkie, Reddit — other people talking about it. That’s much harder to fake.

Make Yourself Worth Recommending

Ask AI the questions your customers actually ask, not your brand name. See who shows up, and why and which sources it’s trusting. Then close the gap: earn real coverage, produce real research, put real experts in front of real journalists, make the evidence machine-legible via SEO/AEO. Measure AI mentions, citations, sentiment and referral traffic. Also check whether AI can even crawl your site. Adobe found big chunks of major retail sites aren’t fully accessible to AI systems at all!

Stop Trying to Outwit the Robot

Agent Ads may mature into a legitimate format. Bit so far, there’s no evidence that buying one buys a recommendation. And Perplexity’s reaction proves the machine gets a vote. AI companies depend on people trusting their answers. The moment those answers look bought, trust erodes fast.

Which brings us to an irritatingly old-fashioned conclusion: build something good, earn credible coverage, make it easy for AI to find. Do the real marketing work instead of hunting for the cheat code. We’ve built extraordinarily sophisticated AI, and our first instinct is apparently to see if we can fool it. 

Maybe the robots aren’t the ones who need to get smarter.

As we preach in our PR for Robots webinars, it isn’t about tricking AI into thinking your brand matters. It’s about doing the work that gives AI enough credible evidence to know that it does.

Sources: The Business Insider (Aug 2026); Muck Rack, What Is AI Reading? (2026); Adobe Digital Insights (2026); Similarweb (2026); Semrush AI Visibility Index (2026); Ahrefs (2026); Google Search Central.

This one’s for anyone who’s ever watched a brilliant campaign work exactly as planned…and ruin the thing it was promoting.

Every event marketer is trained to do the same three things: get bigger, get louder, get easier to reach. That’s sound advice, until we realize the brand’s entire appeal was being smaller, quieter and slightly hard to get into. 

Case in point: Wimbledon. The Guardian recently reported on the tournament’s very deliberate courtship of younger fans, international visitors and a small army of influencers. And it’s working beautifully. The crowd is younger, more global, and producing the kind of content usually reserved for fashion week and infinity pools.

Excellent marketing. Almost too excellent.

Because Wimbledon now risks becoming less a tennis championship than a very photogenic backdrop. The outfits, the flowers, the celebrity sightings and the Instagram proof-of-attendance are starting to out-compete the actual game of elite athletes hitting a ball across a net. That’s the strange thing about overmarketing: it never looks like failure. It looks like a longer queue.

Anyone for Content?

Wimbledon hasn’t made a marketing mistake. It’s done exactly what every brand is told to do: widen access, chase youth, recruit creators, generate content, turn tradition into “experience.” The trouble is that Wimbledon’s whole mystique was built on being not quite accessible. The ballot, the queue, the all-white dress code, the hush, the vague sense that eating a strawberry incorrectly might get you removed. None of that was incidental. That was the show.

Some brands genuinely improve when friction disappears: Amazon should make buying batteries effortless, Uber should make finding a car effortless. But luxury, sport, hospitality and culture often run on the opposite fuel: scarcity, ritual, anticipation. Not every velvet rope is an inefficiency waiting to be optimized away. Sometimes the rope is the product.

Break Point at 29,000 Feet

Everest is the purest version of this. It once stood for isolation and magnificent human stubbornness. Now, photos periodically show a line of climbers queuing near the summit, as if the world’s most dangerous check-in is now its most famous photo wall.

The mountain hasn’t gotten shorter, and the death zone hasn’t gotten friendlier. It’s just become easier for more people to believe that money, gear and enthusiasm are a substitute for experience. Nepal has been weighing tighter permit rules after years of concern about overcrowding and under-qualified climbers. Everest doesn’t need more brand awareness. It needs a bouncer.

Most businesses are lucky: when marketing outpaces supply, customers just complain. On Everest, they can freeze to death waiting for the influencer ahead of them to finish the photo.

Foot-Fault Tourism

Cities have run the same play at scale. Venice and Barcelona spent decades selling beauty, authenticity and “real local life”. That meant millions of people arrived looking for the exact same undiscovered café and secret street. Which is its own small paradox: the neighborhood bakery becomes a backdrop, apartments become short-lets, residents move out, and the destination campaign keeps advertising the authenticity its own success has been quietly evicting.

Across Europe, the backlash has arrived as visitor fees, entry restrictions and, in Barcelona’s case, residents armed with water pistols — not, one assumes, in the original brand guidelines. Arrival numbers are easy to count and great for a slide deck. Carrying capacity, resident sentiment and whether locals still want you there are less exciting, and far more important. A fully booked city isn’t automatically a healthy one.

TikTok, Set, Match

Social media has industrialized the destruction of the hidden gem. Every tiny restaurant, secluded beach and quiet village is now content-in-waiting, ready to be filmed, captioned “literally nobody knows about this place,” and shown to 800,000 people who will all arrive by Friday. “Hidden gem” has quietly become less of a recommendation and more of an evacuation notice.

The pattern is efficient: a creator finds a small restaurant, a crowd arrives for the same sandwich, staff get overwhelmed, regulars disappear, service slips and the reviews complain the place no longer resembles the video that caused the stampede. The algorithm has no concept of capacity. It doesn’t know the restaurant has fourteen tables. It detected engagement. Job done.

Advantage: Everyone

Luxury brands live inside a genuinely funny contradiction: they want to be exclusive and they want to grow, and those two goals actively hate each other. So, the industry expands stores, launches entry-level products, raises prices, seeds influencers and puts the logo everywhere from airport lounges to brunch. The once-rare object is now in every mall, feed and aspirational bathroom mirror. It’s still expensive. It no longer feels rare. And expensive-but-common is a precarious place for a luxury brand to stand.

Bain reports the global luxury consumer base shrank from roughly 400 million in 2022 to about 340 million in 2025, with active shoppers falling even faster. There are several causes, but nonstop exposure paired with nonstop price hikes hasn’t exactly helped anyone feel special. Growth can raise sales while quietly lowering desire — a genuinely awkward slide to present at the quarterly review.

Centre Court, No Centre

Festivals have gone furthest down this road. Coachella is now as famous for outfits, brand activations and sponsored posts as for whoever’s actually on stage. People don’t just attend. They instead document themselves attending, next to other people documenting themselves attending, occasionally blocking the view of people who mistakenly came for the music.

The experience becomes scenery for proof of the experience. The concert becomes the background. The restaurant becomes the studio. Wimbledon becomes the outfit. Eventually the event stops being something people live through and becomes something they rent, wristband included.

The Overmarketing Tiebreaker

Brands rarely spot overmarketing early, because the first numbers look fantastic — reach up, footfall up, mentions up — while operations quietly barricades the door. The real warning signs show up elsewhere: regulars feel crowded out, quality slips, queues become the main event and people seem more interested in proving they showed up than in actually being there.

The right question was never “how many people can we attract.” It’s “how many people can we serve brilliantly without wrecking what attracted them.” For some brands that means maximum reach. For others it means optimum reach — timed entry, memberships, limited releases or deliberately steering demand toward quieter moments. And it means measuring more than awareness and revenue: repeat satisfaction, service consistency, staff strain, community sentiment, and whether the experience still resembles the promise.

A packed house can be a triumph. It can also be a fire-code violation with excellent engagement numbers.

Serve and Protect

Wimbledon’s real challenge was never attracting more people. It’s protecting the ritual and slightly absurd dignity that made people want in to begin with. The queue, the ballot, the strawberries, the near-comical formality — these aren’t outdated obstacles waiting for a consultant to streamline them. They’re the brand.

Marketing has spent decades tearing down barriers, and usually that’s the right instinct. But not always. Sometimes the queue builds anticipation. Sometimes the invitation creates belonging. Sometimes the secret is the whole point.

Sources: The Guardian, “How social media is changing Wimbledon from eminent tournament to tourist event,” July 2026, Bain & Company Global Luxury Market Study. 

Every time we run one of our PR for Robots webinars, we spend most of the time on the same unglamorous point: third-party endorsement is what the machines actually believe. Earned media. Independent citation. Other people’s sentences about you. This is the point of our thesis and why PR has emerged as a more vital discipline in the age of AI.

And every time, somebody asks the same question. So what about our own channels?

Awkward pause. Because the honest answer is: to a robot, your social feed, influencer campaign, content strategy, brand campaign, etc. is merely decoration.

But to a human being, those channels are still where persuasion happens. Which leaves us here — marketing has quietly bifurcated. We now market to two brains. One organic, one synthetic. One wants to be moved, the other wants to be sure. Both have to be served, and only one of them decides whether the other ever shows up.

One important battleground is the website.

For thirty years we pared websites down. Less copy. Bigger hero. Five words and a button. Beautiful. And now the fastest-growing reader on earth turns up, doesn’t even have eye and wants everything we deleted.

The mistake would be to overcorrect and build for the robots alone — a spreadsheet with a logo. The purpose of a website hasn’t changed at all: credibility and information. We just have to fashion it for two audiences at once.

The First Visitor to Your Website Isn’t Human Anymore

Roughly 58.5% of US Google searches already end without a single click, and when an AI Overview appears the zero-click rate reaches about 83%. In May 2026, Google AI Mode became the default experience for over a billion monthly users, which means the synthesized answer, not the ten blue links, is now the front door to your business.

ChatGPT, Gemini, Claude and Perplexity don’t admire your clever tagline or beautiful images. They assemble evidence, deliver a verdict and move on. Your homepage was one exhibit among many.

Your Marketing Has Split in Two

Semrush’s 2026 AI Visibility Index analyzed 126 million US prompts and reached a conclusion that should be tacked above every creative director’s desk: brands now compete not only to be found, but to be accurately understood. Their own framing for the era is “building brands for humans and machines.”

Which is precisely the problem, because the two want opposite things.

Humans respond to emotion, story, surprise and personality. Machines respond to entities, structure, dates, definitions and evidence. 

Robots have joined the target-audience roster alongside customers, media and stakeholders. They are also the only audience most marketers have never surveyed: 45% of marketing leaders admit they cannot measure how their brand appears in AI answers at all.

Most Of You Never Finished the Last Homework Assignment

Two decades into SEO, a startling number of company websites still can’t be crawled properly. Only 43% of marketers are actively implementing GEO — up from near zero a year ago, which tells you how new the panic is. And only 14% track whether AI mentions them at all, meaning 86% are being described to buyers daily by a system they have never once checked.

Nobody gets to skip a grade, either. Google’s AI Mode draws exclusively from Google’s own index, so a page that doesn’t rank can’t be cited. GEO is built on SEO, not instead of it. 

So what actually changed? SEO was a competition for position. You wanted to be the first blue link, and everything — keywords, backlinks, page speed — served that one ambition. GEO is a competition for inclusion. There is no position to win, because there’s no list; there’s a paragraph, and you’re either in it or you don’t exist. SEO asked “can Google find me?” GEO asks “can a machine understand me well enough to recommend me?” The first was a popularity contest. The second is a reference check.

The distinction that matters most is the unit being judged. SEO judged the page. GEO judges the sentence. AI doesn’t retrieve pages; it retrieves passages. One study of 153,000 citations found the average cited sentence ran about nine words, and that roughly 77% of cited URLs weren’t in the organic top ten at all — which means the thing being judged isn’t your site. It’s your sentences, one at a time, out of context, without your logo anywhere near them.

Every sentence you publish is now traveling alone.

Why Websites Suddenly Matter Again

Search engines once merely helped humans find your site. AI reads your site so humans never have to. That inverts the entire purpose of the page and, inconveniently, makes it more important rather than less.

The traffic proves it. Generative AI platforms grew referrals roughly 796% year over year across 2.3 billion sessions, and those visitors convert several times better than classic organic. Adobe’s data shows AI traffic to US retail sites up 1,324% and to travel sites up 2,215% between October 2024 and May 2026.

The Great Website Irony

Princeton’s GEO research found structured, fact-dense content achieves up to 40% higher visibility in AI answers than prose-heavy pages covering identical ground.

Everything we spent ten years deleting — the FAQ, the leadership bios, the specifications, the dull page nobody visited — turns out to be exactly what the machine came for. We stripped the house to the studs for the sake of elegance, and the new buyer wanted the wiring diagram.

Nobody has ever called Wikipedia beautiful. Yet ChatGPT cites an average of fifteen sources per response and leans on Wikipedia and Reddit constantly. Structure beats decoration, which is an unpleasant thing to tell a creative director. I’ve been one for decades. I’m telling you anyway.

The Website Becomes the Source of Truth

Consistency is the whole game. AI cross-checks your description of yourself against every other channel it can reach, which gives the website a new job, and a better one. It is the only channel you fully own, so it should hold the canonical vocabulary: the founder, the expertise, the locations, the services, the awards. Fix the language there, then make LinkedIn, Crunchbase, the directories and every press release echo it verbatim. One dictionary. Every channel spelling your name the same way.

Not to get too technical, but it’s important to understand that Schema.org markup, meanwhile, has graduated from SEO housekeeping to AI infrastructure. Organization, Person, Article and FAQPage do the identity work, the grammar a machine needs to parse you correctly. Google’s own research shows this pays off: pages carrying rich structured data saw an 82% higher click-through rate for Nestlé and a 25% higher click-through rate for Rotten Tomatoes — real brands, real before-and-after numbers, published by Google itself.

PR Becomes the Independent Witness

Here is the part that vindicates our entire industry. Around 85% of brand mentions in AI responses originate from third-party pages rather than owned domains, and roughly 48% of citations come from community platforms. The machine has, without being taught to, arrived at the oldest principle in public relations: consider the source.

Your website makes the claim. The press, the interviews, the reviews and the directories confirm it. Testimony and cross-examination, which is precisely what PR was always for, back when we had to explain it at dinner parties.

It works, too. HubSpot’s 2026 State of AEO report found citations improved 433% when brands actually worked at it, and 44% of buyers have made a purchase from a brand they first met inside an answer engine.

AI Is the Concierge. Humans Are the Guests.

The concierge decides whether to recommend a restaurant. The guest decides whether to book it. A site that flatters only one of them loses.

Look at what that already does to a category. In travel, Expedia, Booking.com, The Ritz-Carlton, Marriott Bonvoy and Hotels.com already dominate what AI recommends without being asked, while most other brands in the same category barely get mentioned at all. Citation share is concentrating fast, and it’s concentrating in a handful of names per category.

And the mechanics favor the few. When an AI Overview shows up, a 2026 field study found the whole query gets 38% fewer clicks overall — but the sources AI actually cites get 35% more clicks than everyone else on that same search. So the total pie shrinks, and whoever gets cited takes a bigger slice of what’s left. There’s no safe, comfortable middle ground anymore — you’re either the source AI is pointing to, or you’re fighting over table scraps.

The Next Great Websites Will Win Both Brains

The developer world moved first, as it always does. Stripe, Cloudflare, Vercel, Anthropic, Coinbase, Supabase, ElevenLabs and Pinecone all publish llms.txt files — plain-language summaries of themselves, written for AI to read. Worth a caveat: their human audience and their machine audience are nearly the same person, a developer, so this is evidence of intent more than proof anyone has solved the harder version of the problem, where the two audiences don’t overlap at all. Adobe now sells an LLM Optimizer to enterprises doing the same thing at scale. Nobody in that list stopped caring what their site looks like. They simply stopped assuming a human would be the one reading it.

And this is one job, not two departments. Organizations that unify SEO, AI visibility and communications report gains at 81%, against 36% for those running them in silos. That’s why Rose promotes an integrated approach. The brief hasn’t split. The audience has. 

So…

Delight the human, inform the machine, and do both in the same sentence. Your site stops being a brochure and becomes evidence, which, if you think about it, is what the good ones always were.

Your website should still move people. But if AI can’t understand it, a growing share of those people will never arrive to be moved.

Sources: SparkToro / Datos, Bain & Company / Dynata, Google I/O 2026, Semrush 2026 AI Visibility Index, GoodFirms, Similarweb, OrganiKPI, Adobe, Media Copilot, Princeton University, Walker Sands, SE Ranking / ALLMO

Heard the one about the brand that outsourced its own personality to AI?

I just read coverage on Tourism Malaysia’s Citrawarna 2026 promo — the one where a government tourism board generated an entire cultural festival with AI instead of pointing a camera at the real cultural assets all around it. AI dancers. AI wood carvers. An AI version of a public square that already exists and was, presumably, available for filming. They even managed to mirror-image their own national flag.

I read this the way you’d read about a restaurant hanging a stock photo of a steak on the wall instead of cooking one and taking a picture of it. It’s not that the photo looks bad. It’s that it looks inauthentic and people notice. (Yes, I know firsthand how much more difficult/expensive it is to take a great food shot than to use stock or AI.  But just go with me on this.)

Countries go to great lengths to preserve and promote their unique culture. It’s the most personal and authentic thing about them. So…WTF were they thinking?

Nobody Asked the Dancers

The clip went up on June 22, featuring the festival’s cuddly mascots, Wira and Manja, walking through an AI-rendered version of Dataran Merdeka, Malaysia’s storied, flag-raising Independence Square, alongside AI depictions of traditional dancers, communities in traditional dress and hand-stretched roti canai flatbread being made. It came down two days later. No statement, no explanation. Just gone, the way something disappears when the people who approved it would rather not discuss why.

One well-known Malaysian comedian put it plainly online: the country has actual cultural dancers, actual food, actual scenery and actual creative people who could have made this. What’s unsettling isn’t that a cost-saving call got made. It’s that someone looked at the AI cut and decided it was good enough to publish.

And as I always say: stock images create stock brands. (Ok. I don’t always say that. But it does sound like something I would say.) 

Nobody Rejected the Tool. They Rejected Being Cut from the Credit.

Here’s the part that should change how you think about this. Social-listening data on the backlash found that 84.6% of the negative sentiment was about authenticity — not opposition to AI as a technology. 

That distinction matters more than the outrage cycle around it. Audiences aren’t rejecting AI. They’re rejecting being told a fake thing is the real thing and rejecting brands that skip the people who’d have made it real.

This isn’t just a Malaysia problem. It’s a cautionary tale for us all.

• 87% of consumers say the best advertising still requires a human touch, and 70% say they can tell something’s missing from an AI ad, without being told why. Audiences don’t need a disclosure label. They just feel it.
• 71% of Gen Z and Millennial consumers now believe they’ve seen an ad made with AI, up from 54% just two years ago. That’s not a niche skill anymore. That’s most of your audience, watching closer than you think.
• Only 13% of consumers say they completely trust AI, full stop, meaning every AI-made campaign starts the relationship already in the red.

It Doesn’t Know Why the Real Dancer Matters

AI is genuinely good at the mechanical stuff like rendering, resizing, filling a brief exactly as written. What it can’t do is feel embarrassed that the dancer isn’t real or come close to understanding what it means to a culture to be represented by something that was never asked to show up. It executes. It doesn’t wince.

Two examples of the same failure in different outfits:

• Meta had to pull an Instagram AI feature within days of launch after it let users generate images from real people’s public photos without asking. Same root problem as Citrawarna: using someone’s likeness, or someone’s culture, without inviting them into the decision.
• Duolingo‘s CEO announced the company was going “AI-first” — cutting contractors, leaning harder on AI — and the backlash was swift: boycott threats, thousands of TikTok unfollows, growth slowing from 60% to 40% YoY. He later admitted he hadn’t explained what any of it meant for users. AI wasn’t new to Duolingo — they’d used it for years. What was new was announcing a decision about people’s future without a word to the people watching.

The technology isn’t the risk. The failure to understand that a brand is a relationship that has two very important and often distinct points of view, as well as the absence of a human raising their hand and asking “should we?” is the risk.

Consistency Is the Whole Point

Trust now shapes 92% of global buying decisions. And brands in the top trust quartile earn a 31% revenue premium over everyone else. Distinctiveness is what earns that trust — the specific, textured, unmistakably-you details that make a brand recognizable at a glance. That’s exactly what gets sanded off when AI fills in the gaps nobody double-checked.  In Tourism Malaysia‘s case, that was the foam on the tarik tea as its poured between two cups, or the woven-leaf texture of ketupat rice cakes steamed inside palm-leaf pouches. Small details. Whole brand identity made generic.

And this compounds, because most brands aren’t being honest about how often it’s happening. Distrust of AI content roughly doubled in twelve months, while only 20% of brands always disclose when they’ve used it. That gap between what audiences want to know and what brands actually tell them is where the damage sits and grows.

What This Actually Means for Your Brand

Strip away the anecdotes and here’s what’s left — the key questions worth asking before AI touches anything customer-facing:

• Does this replace a person we could have credited instead?
• Have we disclosed that this is AI-made, proactively, before someone asks?
• If this claims to represent real people or a real culture, were real people in the room making it and not just approving it after the fact?
• Who in the loop has zero stake in the campaign and gets to say “this feels off” before it goes live?
• Is AI accelerating a decision a human already made, or making the decision for us?
• What should AI never ever be allowed to touch?

Don’t Become a Stock Brand

A stock photo isn’t wrong. It’s technically fine but emotionally empty, and it says nothing about the actual thing it’s supposed to represent. That’s what’s on offer every time a brand lets AI fill in the details nobody bothered to check — a technically competent version of yourself that could belong to anyone.

AI will never be embarrassed on your behalf. That’s still your job. Don’t let it turn you into stock.

Sources: Marketing-Interactive, July 2026, Canva, State of Marketing & AI Report, 2026, IAB, The AI Ad Gap Widens, 2026, Klaviyo, AI Consumer Trends Report, 2026, PwC, Consumer Intelligence Series, 2026, Fractl, AI Search Consumer Trust Study, 2026, TechCrunch, July 2026, Forbes, November 2025

If you’re into stories about how AI messes things up for marketers, you’re gonna love this.  

I just read a Business Insider piece on how Meta’s AI ad tools keep tripping over their own synthetic feet. 

Picture this. A menswear brand’s model got replaced by someone’s grandmother. A shoe company’s model ended up with a leg bent in a direction human legs don’t bend. An e-bike ad started talking about trunk space, which is a feature bikes famously do not have. 

Oh sure, we’re all familiar with these AI image distortions.  But the difference here is that this stuff actually ran!

I read this the way you’d read a police report about your own house being redecorated while you were at work, by a contractor you never hired, who let themselves in through a window they insist was already open.

Here’s what I can’t get past. AI still can’t be trusted anywhere near creative judgment. It’s a tool, not a colleague, and nobody promoted it to art director while the rest of us were asleep. Meta doesn’t much care whose brand gets mangled in the process, as long as the ad spend clears, because it turns out refunds cost more than reputations, and Meta has done the math on which one it’s willing to pay for. 

And any agency letting these features run on autopilot without checking the output isn’t managing your account. They’re collecting the fee and hoping you don’t read Business Insider.

Nobody Asked, Nobody Told

Quick primer, for anyone who doesn’t live inside Ads Manager. When a brand runs an ad on Meta, they don’t just upload a picture and a headline and walk away. They can turn on a suite of tools called Advantage+, Meta’s AI system that’s supposed to automatically improve the ad while it’s running live: testing different crops, swapping headlines, adjusting who sees it, all in the name of performance. In theory, it’s a machine quietly optimizing your ad. In practice, for a growing number of advertisers, it’s a machine quietly remaking your ad, and not always for the better.

True Classic, a menswear brand that sells almost exclusively to men 30-45, found this out the hard way. They had a top-performing ad running with their usual model, a fit millennial guy in a matching fleece set, exactly who their customers want to see themselves as. Advantage+ decided it had a better idea and swapped him out for a smiling grandmother sitting in an armchair. Not a subtle A/B test. A total cast change, mid-campaign, with zero heads-up. It ran for days before customers started asking the brand what, exactly, was going on. True Classic hadn’t approved the change. They didn’t know it happened until their own customers told them.

It gets better, or worse, depending on your appetite for irony. The CEO of Flat Circle, an agency that manages roughly $100 million a year in Meta ad spend, says the platform’s auto-edit settings re-enable themselves after being manually turned off. Not once. Repeatedly. His team now checks the settings several times a week, permanently, like they’re monitoring a patient who keeps pulling out their own IV.

Who’s Actually Driving

The old chain of command was simple. Agency creates, client approves, platform delivers. Everyone knew their job, and nobody’s job was “surprise.”

The new chain of command is: platform edits, platform decides, and tells you afterward, if you’re lucky. The approval step didn’t get faster. It got skipped.

And this is happening at genuine scale. Global ad spend crosses $1 trillion in 2026 for the first time, with AI-generated creative now approaching 40% of all digital video ads, up from almost nothing two years ago.

Consistency Was the Whole Business Model

Consistency isn’t a design preference, it’s the entire business model. Brands like Rolex, LEGO and Coca-Cola haven’t changed their core visual identity in decades, because sameness is what makes them instantly recognizable at a glance, anywhere in the world. Nobody has to think twice about what they’re looking at. That’s not an accident, that’s decades of discipline.

Research from Lucidpress/Marq, a brand-management research firm, found companies with consistent brand presentation across channels average 10-20% revenue growth, with the most disciplined seeing up to 33%. That’s not a soft, feel-good branding stat. That’s real money, sitting on the table, waiting for a company to either earn it or let an algorithm give it away.

So, when an algorithm swaps your model, twists a product photo or rewrites your headline, as happened to Kirruna, a European footwear brand whose Meta-generated ad showed a model’s leg bent the wrong way, it’s not just an ugly image. It’s actively working against the thing that makes a brand worth more than its product.

It Doesn’t Know Why the Rule Exists

AI is genuinely good at the mechanical stuff: resizing, translating, testing variants. It’s bad at knowing why a rule exists in the first place, because it was never told the rule was a rule. It just saw a pattern and decided to improve on it.

Same reporting, another case: Meta’s AI generated an ad for Lectric, an e-bike company, and rewrote the copy to describe a car’s trunk space instead of the bike. Nobody approved it. Nobody caught it until it was already live and doing damage.

The pitch for using AI was “less work.” The reality is a new job: someone now has to check every AI-touched asset before it airs, and again after, because sometimes the settings drift back to ON whether you asked them to or not.

What This Actually Means For You

Strip away the anecdotes and this is what’s left: a short list of hard rules and the questions you should be asking before you let any of this near your brand.

An opt-out that doesn’t stay opted out isn’t a bug. Verify it weekly or assume it’s by default, on. Every AI-altered asset that airs unchecked is a small withdrawal from your brand’s recognition account. Before any AI tool touches customer-facing content, ask who’s accountable if it’s wrong in public, and if nobody can answer that in one sentence, you don’t have a policy. If a tool needs a full-time human minder, that’s a headcount cost. 

Here are six questions worth asking before you trust it:

• Which of our tools can alter creative without a human sign-off?
• Who’s personally accountable for approving what AI generates?
• Have we actually confirmed our opt-outs are still on this week?
• Do we audit outputs on a schedule or only after a complaint?
• Who makes the final creative call, a person or a settings menu?
• What should AI never be allowed to touch?

Who’s Watching the Software

For two years the fear was AI replacing creative people. Wrong worry. The real one: who’s watching the software when it starts making decisions?

Sources: Business Insider, July 2026; Crypto Briefing, July 2026; Lucidpress/Marq State of Brand Consistency Report; Dentsu Global Ad Spend Forecast 2026

The UK recently put M&M’s on trial. The popular candy-coated chocolate maker just learned that Britain’s new restrictions on advertising foods high in fat, salt and sugar that came into force on 5 January 2026 has teeth. Before that, the rulings under it have been rare enough that marketers were still waiting to see how far the law would reach. Now they know.

Mars ran a paid Instagram ad for M&M’s. No bag of candy in sight. No bite, no crunch, no product shot. Just two cartoon characters, a round green one and an oval yellow one, standing next to the words “For a truly unique event” and a link to the M&M’s website. Mars called this brand advertising, the kind explicitly exempted from the HFSS rules because it promotes an identity rather than a specific product.

Britain’s Advertising Standards Authority disagreed. Funny enough, the green character survived. The ASA ruled it represents a whole range of M&M’s flavors and colors, not one specific product. The yellow one did not. Its oval shape, the regulator found, identifies Peanut M&M’s specifically, a distinct HFSS product in its own right. Same ad, same campaign, same brand. One mascot cleared, one convicted. 

Read that twice. A cartoon character with a face and arms just got treated as a proxy for a product ingredient list. Brand assets are becoming the product in the eyes of regulators, and this is not confined to confectionery. Similar arguments are already surfacing around alcohol, gambling and vaping, categories where the mascot, the color palette or the typeface can carry as much regulatory weight as the thing it is selling.

A Peanut Walks Into A Courtroom.

The complaint came from Bite Back, the food campaign group founded to push for tighter marketing rules on unhealthy products aimed at children. That is the enforcement model now. Advocacy groups file complaints, regulators adjudicate, and the resulting rulings become precedent that reaches well past the original brand. Consumer groups, environmental organizations and health advocates play the same role in Australia, Europe and North America. The complaint is the trigger. The ruling is the weapon.

Mars was not the only brand in the crosshairs of this particular round of rulings. Morley’s Woking, a London chicken chain, got the same treatment for an Instagram meal deal ad showing burgers, wings and fries. Uber Eats, by contrast, walked away clean on ads featuring Burger King, KFC, Domino’s and Papa Johns, because the ASA decided the food shown was not visually tied to one specific HFSS product. The line between promoting a brand and promoting a product now runs through individual pixels, and it moves case by case.

There is something almost absurd about a cartoon candy having its shape cross-examined by a regulator. Highly stylized characters with arms, legs and personalities, deployed on merchandise, clothing and partnerships that have nothing to do with food, being read as product depictions anyway. Absurd, and also entirely real. Creative symbolism has become regulatory evidence, and the brands that dismiss this as a quirky one-off are the ones most likely to be next.

The Writing’s On The Wrapper.

The UK is not moving alone. Chile restricts cartoon characters on unhealthy food packaging and advertising. Mexico has front-of-pack warning labels and advertising reform. Quebec has banned commercial advertising directed at children for decades. Australia is cracking down on greenwashing claims. The US Federal Trade Commission continues to pursue deceptive influencer marketing. The UAE and Saudi Arabia have both tightened rules on influencer disclosure, health claims and misleading advertising. The EU keeps pushing digital advertising transparency and platform accountability further.

Different legal systems, different politics, different starting points, same direction of travel: health claims, environmental claims, influencer marketing, advertising to children, financial promotions, misleading advertising. All converging toward the same demand, that a brand be able to prove what it says and defend what it shows.

The scale of what is driving this is not small. More than one billion people worldwide now live with obesity, according to the WHO and the Lancet’s 2024 global analysis, a milestone reached faster than the World Health Organization’s own projections anticipated. When the underlying public health numbers move like that, advertising law moves with them, and it does not move back.

More Than A Mouthful.

Marketers used to build for three audiences. Consumers, competitors, media. That list now runs longer: regulators, advocacy organizations, politicians, activists, courts, platforms. Every one of them can read a campaign and reach a different verdict than the one your creative team intended.

Reputation and compliance are no longer separate departments. The global influencer marketing industry is now worth more than $32 billion a year, per Influencer Marketing Hub’s 2026 figures, a channel built almost entirely on personal trust and light-touch disclosure. Regulatory scrutiny of environmental claims keeps climbing across Europe, Australia and North America. Enforcement actions involving influencers, greenwashing and undisclosed endorsements keep landing.

AI Has Turned Up the Heat.

The old enforcement model ran on complaints. Someone had to notice, someone had to care, someone had to file. That is changing. AI is increasing the scale and speed of advertising monitoring, which means regulators depend less on the public showing up with pitchforks and more on systems that scan everything, all the time. That does not necessarily mean more regulations. It means more routine investigations under the regulations already on the books, faster identification of misleading claims, undisclosed influencers, inconsistent messaging across markets and cross-border violations that used to slip through because nobody was watching every jurisdiction at once.

Read The Label Before You Print It.

Legal and compliance review needs to happen earlier, not as a final checkpoint before launch but as part of the creative process itself. Brand assets, symbols, implied claims and disclosures all deserve the scrutiny that used to be reserved for the product claims themselves. Before a campaign ships, four questions are worth asking every time.

1. Could a regulator interpret this differently than we intend. 
2. Could another market reach a different conclusion than this one. 
3. Could a brand asset itself become the product, the way an oval yellow candy just did.
4. Could this campaign become tomorrow’s precedent, the case study some other brand’s compliance team cites two years from now.

The M&M’s ruling joins a growing pile of reminders. Greenwashing investigations keep hitting major global brands. Influencer campaigns across the GCC and beyond face rising scrutiny over disclosure. None of these started as headline news. They started as one ad, one complaint, one ruling that nobody outside the legal team read closely enough at the time.

The Sweetest Lesson Of All.

Treat the M&M’s ruling as an early indicator, not an isolated curiosity. The global regulatory environment is tightening steadily, with legal scrutiny of brands rising across industries and jurisdictions, and with different political and legal systems converging, oddly, on the same stricter standards.

The M&M ruling is not really about candy. It is an early warning that marketers everywhere are entering a more heavily regulated era in which distinctive brand assets, once considered marketing strengths, are increasingly becoming legal liabilities.

For over forty years I get paid to find the words that make a product sound irresistible. Biggest, smallest, fastest, new & improved. Also: light, diet, low-fat, sugar-free, natural, organic. And now: sustainable, carbon-neutral, and, currently trending, AI-powered. Every generation of marketers invents its own magic words because every generation of customers falls for different bait. There’s nothing sinister in that. It’s just the business.

What’s changed is that nobody’s treating those words as just marketing anymore. They’re treating them as evidence. And evidence, unlike copy, has to hold up in court.

Recycled, Reused, Re-Litigated

In June 2026 the UK’s Advertising Standards Authority banned ads from Adidas, Calvin Klein and Uniqlo for using the word “recycled” in ways they couldn’t back up. 

Adidas advertised a “recycled shoe range” it turned out didn’t exist. Uniqlo’s fleece was recycled in the parts you could see and not recycled in the parts you couldn’t. Calvin Klein got dinged for implying an entire collection was made from preferred materials when the real number ranged from 20% to 100%. 

Six months earlier the same regulator had done the exact same thing to Nike, Lacoste and Superdry. Six brands. Two rulings. One word. The lesson isn’t about fashion. It’s about the word “recycled” no longer surviving close contact with a lawyer.

The Robot Reads the Fine Print Now

Here’s the part that really changed the game. The ASA didn’t catch any of this from a customer complaint. It caught it with an AI system called Active Ad Monitoring, which scanned nearly 60 million ads in 2025 alone and now accounts for close to half the regulator’s total workload. That intelligence helped resolve more than 40,000 complaints covering over 25,000 ads, resulting in more than 22,000 amendments or withdrawals. 

The old model was: someone gets annoyed, someone complains, someone investigates. The new model is: the machine reads every ad on the internet before breakfast and flags the liars. For decades marketers worried about focus groups. Now we should be worrying about a bot with better reading comprehension than all compliance departments put together.

Every Brand Should Take Notice

Lufthansa, Air France-KLM, Etihad and Virgin Atlantic have all had ads banned for implying that flying with them was somehow kinder to the planet than flying with anyone else. None of these airlines were faking their sustainability investments. Sustainable aviation fuel is real, offset programs are real. What wasn’t real was the insinuation that any of it added up to guilt-free air travel. Progress is not the same thing as permission to round up.

Healthcare Suffers the Same Scrutiny, with Sharper Teeth. In early 2026 the FDA sent Novo Nordisk two separate warning letters in the space of a month for misleading ads by Ozempic and Wegovy, their diabetes and chronic weight management drugs. One of them literally implied superiority over other GLP-1 drugs with no data to support it. Weeks later the agency issued warning letters to 30 telehealth companies for misleading marketing of compounded GLP-1 products, part of a six-month enforcement wave that produced more warning letters than the entire previous decade combined. Meanwhile in France, regulators fined Novo Nordisk close to €1.8 million and Eli Lilly over six figures for “disease awareness” campaigns that never named a drug but, according to the regulator, didn’t need to. Everyone in the room already knew which drug they meant.

Even the Category Name Is a Claim Now

Just to prove the trend has no borders and no mercy, India’s food safety regulator recently issued notices to Red Bull, Sting, Monster and several other brands for calling themselves “energy drinks” (a category that, as it turns out, doesn’t officially exist under Indian food law). Marketers get so comfortable with category language that we stop noticing it’s actually a claim. Apparently regulators noticed for us.

Patriotism, Audited

The FTC launched a “Made in the USA” enforcement sweep in April 2026, settling with sellers of flags, entertainment systems, and footwear for claiming products were American-made when they were partly stitched together in the Dominican Republic and Brazil. One company advertised boots as “handcrafted 100%” domestically while sourcing components from two other countries. “Made in,” “Built in,” and “Designed in” sound interchangeable in a brainstorm. They are not interchangeable to a federal regulator with a customs manifest.

The Cancel Button Is Also Copy

And it’s not just the ad anymore. The FTC’s $2.5 billion settlement with Amazon ($1 billion in penalties, $1.5 billion back to roughly 35 million affected customers) was about Prime enrollment and cancellation flows, not a single banner ad. The checkout screen, the “are you sure?” pop-up, the maze between “cancel” and “confirm cancel,” all of it is now marketing communication, and all of it is now litigable.

The AI Doesn’t Forget What You Deleted

Then there’s the part that actually raises the stakes. BCG found that shopping-related generative AI use grew 35% between February and November 2025. Capgemini found that a quarter of consumers already used generative AI shopping tools in 2025, with another 31% planning to. That matters because an AI assistant doesn’t just repeat your tagline back to you: it cross-references it. It notices when your sustainability page says one thing and your annual report says another. It remembers the claim you quietly walked back last year. Marketing used to compete for attention. It’s now competing for something much harder to fake: consistency across every document you’ve ever published.

Trust Is the New Adjective

Edelman’s latest Trust Barometer puts it plainly: 88% of consumers say trust is an important factor in deciding whether to buy from a brand. That’s not new information: trust has mattered since the first merchant sold the first not so fresh fish. What’s new is how fast it can now be tested, by a regulator, a journalist, a competitor or a chatbot, often within the same week the campaign launches.

What This Actually Means for the Work

The job hasn’t gotten smaller. It’s gotten more interesting. The old question was “how can we make this sound better?” The new question is “how can we make this true enough that nobody, not a regulator, not a rival, not an algorithm, can take it apart?” That’s a harder brief. It’s also, frankly, the brief we should have been writing to all along.

What Marketers Should Actually Do About It

• Treat every claim as something that will eventually be read by a regulator, a competitor, a journalist, and an AI model, because it will.
• Replace the adjective with the number. “Sustainable” is an opinion. “40% recycled polyester, independently certified” is a fact.
• Assume nothing about a fashionable word’s meaning. “Recycled,” “natural,” and “AI-powered” all mean whatever a regulator decides they mean this year.
• Remember that the checkout page, the cancellation flow, and the subscription terms are marketing copy too, and they get read in court.
• Build the proof before you write the headline, not after someone asks for it.
• Get legal and compliance into the room during the creative process, not after the campaign’s already shot.
• Write claims that will still be true in five years, not just until the next sales meeting.
• Stop asking how good your copy sounds. Start asking how fast someone else could prove it wrong.

Sources: European Commission, Green Claims / greenwashing study; UK Competition & Markets Authority; UK Advertising Standards Authority (ASA/CAP), including the ASA and CAP Annual Report 2025 and Active Ad Monitoring briefing; U.S. Federal Trade Commission; U.S. Food & Drug Administration; France ANSM (Agence Nationale de Sécurité du Médicament et des Produits de Santé); India FSSAI (Food Safety and Standards Authority of India); Boston Consulting Group (BCG); Capgemini Research Institute; Edelman Trust Barometer

I was reading a Fast Company article about Pirate’s Booty, the puffed rice-and-corn snack with the wonderfully ridiculous name, and its new rebrand. It reminded me that marketers never seem to learn one basic lesson: a successful brand can become sacred to the people who buy it. Maybe there were too many young mateys in the brand meeting and not enough weathered pirates who knew which treasure not to bury.

Pirate’s Booty, now part of The Hershey Co., was created in 1987 by entrepreneur Robert Ehrlich as a healthier alternative to other kids’ snacks.  It’s one of those brands whose charm has always lived partly in its own silliness. It is a cheesy puffed snack with a pirate, a treasure-chest and a name that sounds as if it was approved by a seven-year-old.

The original pirate had an expression that seemed to say, “Hi! I’m not wearing pants,” while the parrot had an expression that seemed to say, “Hi! He’s definitely not wearing pants.” That was part of the charm. Nobody was looking to Pirate’s Booty for maritime trouser compliance.

Now the artwork has moved from a flatter, more hand-drawn 2D style to a more dimensional, animated-looking 3D world. Yes, the new pirate is clearly wearing pants, in case you were wondering. But what I really want to know is why no one is talking about the fact that his eye patch appears to have changed sides? To me, that should be the major controversy. This poor man has somehow suffered the misfortune of gaining eyesight back in one eye while losing it in the other. Or worse, maybe he never had an eye problem in the first place, and the brand has been living with this lie all along.

My point is that the old packaging had a scruffy, lunchbox, treasure-map quality. The pirate felt like he belonged on a snack bag, not on a brand strategy slide. Then came the redesign: cleaner, smoother, neater, more modern, more polished and more professionally behaved. In other words, exactly the sort of thing that can sometimes make a brand worse while making every individual element look “better.”

When Better Looks Worse

This is where many companies get into trouble. They confuse modernization with cosmetic surgery. They take something familiar, loved and a little odd, send it away for expensive work, then act surprised when it comes back looking younger, tighter and somehow less itself.

A bad rebrand is a lot like the backlash a beloved actress gets after a nip and tuck. The public reaction is rarely, “What marvelous structural refinement.” It’s usually, “What happened to her?” That’s because people don’t fall in love with symmetry. They fall in love with expression, character and familiarity.

Brands work the same way. You can simplify a logo, polish a mascot, update the packaging, refine the colors and improve every visible element, yet still damage the thing people actually liked. The brand comes back from surgery looking more contemporary, but less familiar. Customers notice, even if they don’t describe it in the language of brand equity. They never say, “You damaged a distinctive memory structure.” They say something much more useful: “Why did you ruin it?”

The Customer Owns the Feeling

Marketers should listen to that, because a successful brand is the love child of a company and a customer. The company creates the name, logo, package, mascot, color palette, tone and promise. But customers add the memory. They add the recognition, ritual, habit, trust, nostalgia, loyalty and emotional meaning. The company may own the trademark, but the customer owns the feeling.

That’s why changing a brand isn’t just a design decision. Its relationship surgery. And when companies operate without understanding what customers are emotionally attached to, they shouldn’t be shocked when the patient wakes up angry.

Inside companies, familiar brand assets often start to feel tired. The brand team has seen the logo too many times. The agency wants to make its mark. The new CMO wants proof of momentum. The board wants modernization. Someone says the current identity is not “digital-first,” and before anyone can say “distinctive assets,” a beloved mascot is being quietly made to walk the plank.

But customers aren’t in those meetings. They’re not tired of the brand in the same way the brand team is tired of the brand. They see it for a few seconds on a shelf, sign, app icon, aircraft tail, streaming screen or restaurant wall. Familiarity is not boredom to them. Familiarity is usefulness. This is where companies get dangerously stupid. They look at a familiar symbol and see age. Customers look at the same symbol and see home.

Clean Can Become Empty

Cracker Barrel learned that lesson the hard way. When the company removed the familiar old man from its logo, it may have thought it was simplifying. Many customers saw something else. They saw a brand removing one of the cues that made it feel like Cracker Barrel.

The old man wasn’t elegant. He wasn’t sleek. Exactly. He was rocking chairs, biscuits, road trips, country-store clutter, fake old-timey comfort and the promise that nothing inside would be too surprising. He wasn’t decoration. He was a shortcut to the whole experience.

The backlash became political, cultural and emotional almost immediately, but underneath the noise was a very simple marketing lesson. Customers had decided that the old man belonged to the brand. When the company removed him, customers behaved as if something had been taken from them. Eventually, the company reversed course. That is the thing about brand symbols: from the inside they often look decorative, but from the outside they may be foundational.

Recognition Is Not Decoration

The same mistake can happen with packaging. Tropicana famously replaced its familiar orange-with-a-straw image with a cleaner, more premium-looking carton. In a design presentation, it probably looked fresh. On a supermarket shelf, it became harder to find. The orange wasn’t just a picture. It was a shopping shortcut. It said, “Here I am. I am the one you already know and love” When Tropicana removed it, the brand didn’t simply lose a visual device. It damaged recognition at the exact moment recognition mattered most.

This is why distinctive brand assets matter so much. Logos, colors, mascots, slogans, package shapes and product cues are not merely aesthetic choices. They are commercial infrastructure. They help people recognize, choose and trust a brand without having to think too much about it. One study of brand codes found that only about 15% of tested assets were truly distinctive. That should terrify marketers. If your brand is lucky enough to own something people recognize, why casually sand it down until it looks like everything else?

The lesson isn’t that old is always better. Sometimes old is just old. Sometimes a logo really does need to be improved, a package really does need to be simplified, and a brand really does need to evolve. But there is a difference between removing clutter and removing character. There is a difference between modernizing a brand and making it anonymous.

The Meaning Test Beats the Taste Test

Coca-Cola learned an even bigger version of this lesson all those years ago with New Coke. The company thought it had a product problem. Pepsi was winning taste-test battles. A sweeter formula tested well. So, Coca-Cola changed the formula. On paper, it was rational. In real life, it was madness.

People weren’t just drinking a fizzy brown liquid. They were drinking memory, Americana, habit, family refrigerators, gas stations, ballgames and one of the most loaded pieces of commercial symbolism on earth. Coca-Cola won the taste test and lost the meaning test.

That’s one of the great traps in research. (There’s more than one, but don’t get me started!) It can tell you what people prefer in isolation, but not always what they are emotionally unwilling to lose. When testing a rebrand, companies should not only ask which design looks better. They should ask which design still feels like us, what customers would miss if it disappeared, what people will think we are trying to say, and what meaning we may be accidentally killing.

Even Temporary Changes Touch Memory

Poor Spotify. Its temporary 20th-anniversary app icon was glittery and disco-ball-like. It was not meant to be permanent. It was a campaign flourish. But plenty of users hated it anyway and wanted the familiar green icon back.

That reaction matters precisely because the change was temporary. If people can get irritated by a short-lived icon change, imagine what happens when a company permanently removes a symbol customers use every day to find, trust or emotionally decode a brand.

Global Brands Still Need Roots

British Airways once tried to replace its Union Flag-style tailfins with more international “World Tails” designs. Strategically, the logic was understandable. Airlines are global. Customers are global. Britain itself was changing. The company wanted to look more cosmopolitan. But many customers and critics saw a British airline removing one of the clearest visual signs of Britishness from its aircraft. The lesson isn’t that national symbols are always sacred. The lesson is that rootedness matters. A global brand doesn’t become more global by becoming less itself.

The Backlash Is Now Part of the Launch

The stakes are higher now because trust has become more personal and backlash has become more immediate. Roughly 80% of people say they trust brands they personally use. That’s not abstract corporate goodwill. That’s intimate, practical trust. It’s the trust of “I know this,” “I buy this,” “this has never disappointed me,” “my family uses this,” and “this is part of my life.” So, when a company changes familiar brand symbols, customers can experience it as more than a visual update. The company thinks it changed a logo. The customer thinks it changed the deal.

What’s more, a rebrand no longer launches quietly. Screenshots spread instantly. Critics pounce. Customers mock. AI search summarizes the outrage. Investors notice. Employees cringe. The launch video with the earnest voiceover suddenly looks like evidence at a trial. A rebrand is no longer a controlled announcement. It is an open audition for public ridicule.

The Real Sin Is Disrespect

None of this means brands should never change. Of course they should change. Markets change. Customers change. Channels change. Technology changes. Taste changes. A brand that never evolves eventually becomes a museum piece with a sales target. But smart change makes a brand more itself. Bad change makes a brand more generic. Bad change behaves as if the trademark certificate is the whole relationship.

That’s the arrogance at the heart of so many failed rebrands: believing that because the company owns the mark, it owns the relationship. A successful brand is the love child of a company and a customer. And nobody likes watching one parent give the child a facelift without permission.

What Marketers Should Learn

Before changing a familiar brand asset, marketers should ask what job it actually does. Does it help people recognize you faster? Does it signal trust? Does it carry heritage? Does it make you distinctive? Does it create affection? Does it make customers feel that you are still the brand they chose? If the answer is yes, treat it carefully.

Don’t start with what the brand team is tired of. Start with what customers would miss. Don’t mistake a cleaner logo for a stronger brand. Don’t assume “digital-first” means “personality-last.” Don’t let the strategy deck murder the mascot. And above all, don’t surgically remove the weird little thing people love just because it looks old in a conference room.

Maybe Pirate’s Booty will be fine. Maybe the new pirate will settle in. Maybe kids won’t care. But this lesson is bigger than one pantsless, eyepatch swapping pirate.

When customers love a brand, they don’t only love the product. They love the accumulated meaning around it: the oddness, the shortcuts, the signs, the rituals and the emotional furniture. Change those things carelessly and people may react as if something has been stolen from them. Because, in a way, it has.

Sources: Fast Company — Pirate’s Booty rebrand and packaging update, Reuters — Cracker Barrel logo reversal after backlash, Ipsos / Jones Knowles Ritchie — Distinctive brand assets research, Edelman — 2025 Brand Trust report.

At Cannes Lions this year, WPP Global Chief Creative Officer Rob Reilly warned that human creativity is “under fire” as AI rapidly spreads throughout marketing, advertising and content creation.

Fair enough.

If you wandered through enough conference halls, panel discussions and beachside networking events, you could be forgiven for thinking that artificial intelligence had already become the industry’s new religion. Every agency has an AI strategy. Every consultant has an AI framework. Every platform has an AI solution. Every presentation seems to contain at least three references to AI and one image that was probably generated by AI.

The conversation usually revolves around what AI can do. Can it write? Can it design? Can it create videos? Can it build campaigns? Can it replace agencies? Can it replace marketers? Can it replace creatives?

Interesting questions. But I’m not convinced they are the most important ones.

The more interesting question is this: what happens when everyone suddenly gains access to the same tools? Because that’s what is actually happening.

Much of the current AI conversation assumes that the technology itself will become the source of competitive advantage. History suggests otherwise.

Throughout business history, technology has repeatedly lowered barriers to production. The businesses that ultimately won were rarely the ones that merely adopted the technology. They were the ones that figured out how to use it better than everyone else.

The technology mattered. The people mattered more.

AI appears likely to follow the same pattern. The question is not whether AI will make marketers better. It will. The question is whether it will make them different.

Throughout business history, technology has repeatedly lowered barriers to production. The technology mattered. The people mattered more.

We’ve Seen This Movie Before

When word processors replaced typewriters, writing became dramatically easier. Editing no longer required retyping entire pages. Productivity exploded. Most writers became better writers because they could focus more on ideas and less on mechanics.

Overnight delivery transformed business communication. Documents, contracts and proposals that once took days could suddenly move across the country overnight. Business became faster and more responsive.

The internet accelerated everything again. Information became instantly accessible. Communication became immediate. Publishing became nearly free. Businesses gained access to knowledge, markets and customers that would have been unimaginable only a generation earlier.

Then came desktop publishing, digital photography, blogging platforms, social media and content marketing. Each innovation lowered barriers. Each increased speed. Each increased volume.

In every case, the technology improved performance. The important point is that it improved performance for almost everyone.

Technology rarely creates lasting competitive advantage on its own. Once broadly adopted, it becomes table stakes.

AI may be following the same pattern. The winners may not be the people using AI. The winners may be the people who bring something to AI that everyone else lacks.

For Years We Confused Activity with Talent

Marketing has spent years rewarding production: more content, more campaigns, more channels, more reports, more dashboards, more activity. Technology often allowed average performers to appear exceptional simply by increasing output.

AI accelerates this trend to an entirely new level. Presentations that once took days can be assembled in hours. Articles that once required significant effort can be drafted in minutes. Images that once required designers can be generated almost instantly.

Productivity is exploding, which is wonderful. But also dangerous. Because output and talent are not the same thing. They never were.

The winners may not be the people using AI. The winners may be the people who bring something to AI that everyone else lacks.

AI Does Not Democratize Distinction

For a while, I found myself buying into an argument I was hearing from many AI advocates. Perhaps we humans are arrogant to believe our organic neural networks are somehow special. Why shouldn’t synthetic neural networks, supported by enormous computing power and trained on vast amounts of information, eventually outperform us creatively as well? The more I thought about it, the more reasonable the argument sounded.

Then I realized something important. AI is remarkably good at several aspects of what we commonly call creativity. It excels at recombination, mixing existing ideas in new ways. It excels at pattern completion, extending, refining and riffing on established forms. It excels at volume and speed, generating hundreds of possibilities almost instantly. It excels at style mimicry, reproducing the surface characteristics of human creative work with astonishing accuracy.

Those are not trivial capabilities. In many situations they are extraordinarily valuable.

But they are not the same thing as originality.

What appears to be missing is motivation. Human beings create because they want something. Recognition. Understanding. Revenge. Love. Money. Legacy. Connection. Curiosity. Obsession. Every great creative act begins with a human desire.

AI has no stake in what it creates. It also lacks embodiment. Much of human creativity comes from living inside a body, experiencing time, physical sensation, aging, uncertainty and the uncomfortable awareness that one day we will die. AI does not experience any of those things.

Most importantly, AI struggles with genuine novelty. It works by finding patterns, relationships and possibilities within the universe of existing human knowledge. It can create surprising combinations. It can create beautiful combinations. But it cannot easily step outside the boundaries of human-generated experience.  It may come up with a great idea you never thought about.  But that’s probably because you weren’t very imaginative in the first place.

Above average work looks remarkable to average workers.

Then there is taste. Human beings make creative choices that cost something. They risk rejection. They risk embarrassment. They risk failure. They risk looking foolish. Great creativity often requires vulnerability. But AI literally has no skin in the game.

Which leads to an observation that may sit at the center of this entire debate.

AI creativity is less like an artist and more like a mirror. It reflects, refracts and amplifies human creativity at extraordinary scale and speed. But the light source remains human.

The mirror can produce remarkable images. But it generates no light of its own.

The New Premium Is Human Imagination

The widespread adoption of AI will almost certainly make marketers better. It will make writers better. It will make designers better. It will make strategists better.

The problem is that it will make nearly everyone better at the same time. And when everybody has access to the same amplifier, amplification stops being a differentiator.

Human imagination becomes the differentiator. Judgment becomes the differentiator. Taste becomes the differentiator. Originality becomes the differentiator. The ability to connect unrelated ideas becomes the differentiator. The ability to understand people becomes the differentiator. The ability to spot opportunities before everyone else becomes the differentiator.

In other words, the qualities that have always separated extraordinary marketers from average ones may become even more important. Not less.

AI creativity is less like an artist and more like a mirror.The mirror can produce remarkable images. But it generates no light of its own.

The Gap May Actually Get Bigger

Many people assume AI will level the playing field. The opposite may occur.

As routine production becomes easier, truly talented marketers may pull further ahead. And overreliance on AI will make the untalented fall further behind. When everyone has access to the same tools, differentiation shifts to the quality of thinking behind the work.

The gap between exceptional and average may actually widen. AI may not eliminate competitive advantage. It may amplify it.

The organizations that benefit most from AI may not be the organizations with the most AI. They may be the organizations with the most interesting humans using it.

A Different Kind of Creativity Crisis

Rob Reilly may be right that creativity is under pressure. But perhaps not for the reason most people think.

AI is not threatening creativity because it can create. It is threatening creativity because it is exposing the difference between creation and originality.

The easier it becomes to produce something, the harder it becomes to produce something memorable. The easier it becomes to generate content, the harder it becomes to generate distinction. And distinction is where value lives.

The future may not belong to the marketers who create the most content. It may belong to the marketers who bring the most imagination to the machine. Because the tool keeps getting smarter. 

The challenge is making sure the humans do too.

So this happened.

Starbucks Korea, the South Korean operator of the global coffee chain, launched a promotion for a large drinks tumbler called “Tank.” The campaign was called “Tank Day.” So far, so not so good.

Unfortunately, it launched on May 18, the anniversary of the 1980 Gwangju Uprising, when South Korea’s military government violently suppressed pro-democracy protesters.

You can already hear the brakes failing.

Critics connected “Tank” and “Tank Day” to the military vehicles used in the crackdown. They also connected campaign language about putting the tumbler on the table with a “tak” sound to a phrase associated with the 1987 torture death of student activist Park Jong-chol.

Within hours, the promotion was pulled. The CEO of Starbucks Korea was soon fired. Shinsegae Group, the South Korean retail conglomerate that operates Starbucks Korea, publicly apologized. Sales dropped. Government bodies distanced themselves. Starbucks Korea later announced it would close all stores early on one day in June for mandatory training on history and social sensitivity.

All for a tumbler.

Was the outrage fair. Of course not. Haters gotta hate. But I’ll leave that argument for comment sections, panel shows and people with too much time and very strong views about drinkware.

For marketers, PR heads and CEOs, the lesson is simpler and far less comforting. A campaign does not have to be malicious to become catastrophic. It only has to be interpretable. Starbucks did not need to mean harm for the harm to become real.

“We Didn’t Mean It” Is Not a Crisis Strategy

Starbucks said the incident was unintentional. Shinsegae reportedly found no evidence the campaign was deliberate. I believe that. 

Inside the meeting room, “Tank” probably sounded like product language. Big. Solid. Durable. Heavy-duty. A tumbler with shoulders. “Tank Day” probably sounded like a neat retail mechanic. The launch date may have come from inventory timing, app scheduling or a promotional calendar. The slogan may have been intended as harmless product theater.

Nobody in the room was necessarily thinking, “How do we turn a coffee cup into a national wound?”

But that’s exactly the problem. Campaigns live in the real world, where dates have history, words carry baggage, images have echoes and people are very good at connecting dots. The dots in this case were brutal: Tank. Tank Day. May 18. Gwangju. State violence. A global brand. A slogan that critics tied to another ugly chapter of Korean history.

This is the part marketing people often hate, because we still cling to the fantasy that intent matters most. It does not. Not once activists, consumers, politicians, journalists and bored internet prosecutors concoct a narrative.

The outrage economy is not a courtroom. It does not need proof in any real sense. It needs symbols, timing, screenshots and a villain. And nothing makes a better villain than a large brand explaining that it did not mean to do the thing everyone is furious about.

AI Is Not the Adult in the Room

AI didn’t cause the Starbucks Korea crisis. Reporting says Shinsegae claimed marketers consulted an AI tool for the campaign slogan. Reporting also says some managers who approved the campaign had not opened the attachments showing the materials. This suggests AI may have been somewhere in the chain while humans failed to do the one job AI cannot be trusted to do: understand what an idea may awaken in a specific culture at a specific moment.

And let’s be honest. It is increasingly hard to believe that many campaigns today do not pass through AI somewhere. Naming routes. Copy options. Translations. Search prompts. Social captions. Internal summaries. Research shortcuts. Headline drafts. “Just give me ten options.” “Make it punchier.” “Any issues with this?”

The problem is not that marketers use AI. Everyone is using AI somewhere, in some form, whether officially, quietly or through the intern who is now mysteriously 400% more productive.

The problem is treating AI like judgment.

AI can generate options. It can produce a lot of words very quickly. It can summarize, translate, polish and imitate confidence beautifully. What it cannot reliably do is know when a perfectly ordinary phrase becomes toxic because of history, memory, grief, politics, language, timing or local trauma.

At best, AI is a brilliant intern. It is not the grown-up in the room. Not yet.

And if the grown-ups are approving campaigns without opening the attachments, then the chatbot is not the weakest link in the chain. The chain is the weakest link in the chain.

Context Hijacks Campaigns

Starbucks Korea is not the only brand to discover that a campaign can become about something it was never intended to be about.

Zara, the Spanish fashion retailer owned by Inditex, pulled campaign images in 2023 after critics said visuals of wrapped mannequins, rubble and damaged-looking figures resembled images from the Gaza war. Zara said the campaign had been conceived earlier and was intended to show a sculptor’s studio.

That may be true. It also did not matter. The campaign did not have to be about Gaza to become about Gaza.

This is what marketers need to understand. Context is part of the campaign. If the world changes between concept and launch, the campaign needs to change with it. A fashion shoot can become a war image. A product name can become a historical insult. A slogan can become evidence. A model choice can become a geopolitical statement. A joke can become a boycott.

Localization Is Not Cultural Intelligence

There is a lazy version of global marketing that believes every risk can be solved by localization. Translate the copy. Swap the model. Change the currency. Add local hashtags. Maybe use the right flag emoji if no one gets sleepy. Welcome to our world.

But localization asks whether the campaign fits the market. Cultural intelligence asks what the campaign could awaken.

Swatch, the Swiss watch brand, apologized in 2025 after an ad showing an Asian male model pulling the corners of his eyes triggered backlash in China and calls for boycott. The company removed the material globally and apologized for the distress or misunderstanding caused.

Again, perhaps someone saw fashion mischief. A pose. A face. A moment. A little visual attitude. The audience saw a racist gesture with a long history.

The most dangerous campaigns are often not the ones everyone knows are risky. Those get reviewed, debated, legally cauterized and PR-proofed until they have the pulse of a dentist’s waiting area. The dangerous ones are the ones that look harmless to the people in the room. Which is why the people in the room are often the problem.

“People Are Overreacting” Is Not a Plan

Whenever one of these controversies explodes, someone immediately says, “People are overreacting.” Often, they’re not wrong.

Sometimes outrage is sincere. Sometimes it is performative. Sometimes it is political. Sometimes it is opportunistic. Sometimes it is twelve people on X, three journalists, two politicians and a brand team having a nervous breakdown in real time.

But commercially, that distinction is not as useful as people think. Once the backlash is moving, the brand does not get to stop the train by explaining that the passengers are being unfair.

Adidas, the German sportswear company, revised a 2024 campaign for its SL72 sneaker after criticism over the shoe’s connection to the 1972 Munich Olympics, where Israeli athletes were murdered by Palestinian militants. The campaign featured Bella Hadid, an American model of Palestinian heritage. Adidas said the connections were unintended.

Again, I can see how this happened. Retro shoe. Global model. Fashion image. Campaign calendar. Nice lighting. Someone probably thought it looked great.

But the public saw 1972 Munich, Israel, Palestine, Bella Hadid and a German brand. Suddenly a sneaker campaign had become a political statement.

Whether Adidas meant that statement is almost beside the point.

The Numbers Don’t Care About Your Intent Either

YouGov found that more than seven in ten consumers across 17 international markets said they would boycott a brand if the company or its leaders acted in ways they objected to. In the UAE and Saudi Arabia, YouGov found that 66% of consumers said they had temporarily or permanently boycotted a brand following a scandal.

Kantar’s Brand Inclusion Index found that 75% of consumers say a brand’s diversity and inclusion reputation influences purchase decisions. McKinsey found that 88% of surveyed organizations regularly use AI in at least one business function, with marketing and sales among the most common areas of use.

So we have three forces colliding. Consumers are more willing to punish brands. Brands are producing more content, faster, across more markets. AI is increasingly embedded in the machinery of that production. What could possibly go wrong?

Everything, obviously.

Speed Makes Stupidity Scalable

AI did not invent bad judgment. Marketing departments were perfectly capable of producing stupid ideas long before anyone asked a chatbot to “make this more premium but playful.”

What AI does is make speed feel like intelligence. It can produce a hundred lines in the time it used to take one human to stare out the window and have one useful thought. It can generate campaign names, social captions, slogan routes and localization options with impressive confidence and absolutely no fear of being fired.

And because the output looks finished, everyone starts behaving as if thinking has happened. More output is not more judgment. Sometimes it is just more rope.

Every Campaign Needs a Kill Switch

The answer is not timid marketing. God save us from safe campaigns. Most of them deserve to die quietly in the procurement department.

Brands should still be bold, funny, provocative, fast, strange, distinctive and occasionally dangerous. But bold is not the same as oblivious. Provocative is not the same as handing critics a loaded gun and a Wi-Fi connection.

Every campaign now needs a kill switch. Not a bureaucratic committee that makes the work worse by sanding off anything interesting. Not a late-stage legal review. A real kill switch. Someone with the cultural competence and organizational authority to stop the campaign before it becomes a case study.

That means asking, before launch:

• What date are we launching, and what does that date mean in this market?
• What do the words mean in local slang, politics, history and minority communities?
• Do the visuals resemble current news images, tragedies or conflicts?
• What is the most hostile good-faith interpretation?
• What is the most opportunistic bad-faith interpretation?
• Where did AI touch the work?
• Who reviewed it with actual cultural knowledge?
• Did the approvers actually open the materials?
• What would the headline be if this went wrong?

Those of you who know me, know that last one is my favorite. I often start from the headline and work backwards. It cuts through almost everything. If you cannot bear the headline, do not launch the campaign.

The Adult in the Room

Modern agencies are not just here to make brands more visible. The role of a serious marketing and PR partner is not to say yes more elegantly. It is to ask the irritating question before the public asks it louder. It is to protect the idea from itself. It is to see the second meaning before the internet gives it a lawyer, a hashtag and a moral position.

The adult in the room is not there to ruin the work. The adult is there to stop the work from ruining the company.

Starbucks Korea did not fall to its knees because of a tumbler. It fell because a product name, a campaign name, a date, a slogan, possible AI involvement, unread approvals and national memory collided in public.

Maybe the outrage was excessive. Maybe it was opportunistic. For marketing directors, PR heads and CEOs, that debate is secondary. The outrage economy does not care what you meant.

So. before the campaign goes live, find the adult in the room. And make sure it is not the chatbot.

Sources: Reuters — Starbucks Korea to give staff history training after backlash over marketing campaign; Reuters — Starbucks Korea head fired after “Tank Day” promotion sparks public uproar; Reuters — Shinsegae chairman public apology and reported sales impact after Starbucks Korea backlash; The Guardian — How Starbucks Korea’s “Tank Day” marketing stunt spiralled into mass boycotts; Reuters — Zara pulls campaign after Gaza boycott calls and later expresses regret over “misunderstanding”; The Guardian / People — Adidas apologizes and revises Bella Hadid SL72 campaign linked to 1972 Munich Olympics; Reuters — Swatch apologizes for “slanted eye” ad after backlash in China; YouGov — Global brand boycott survey across 17 markets; YouGov — UAE and KSA consumers boycotting brands following scandals; Kantar — Brand Inclusion Index 2024; McKinsey — The State of AI