Seven in Ten Marketers Have Already Been Burned by AI — But Only a Third Plan to Take It Seriously

I came across a set of numbers recently that stopped me cold.

IAB, together with an AI safety company called Aymara, used their own Insights Engine survey platform (powered by Attest under the hood) to ask 125 U.S. advertising executives one question: how is your AI usage going?

All the respondents came from companies with 50 or more employees, and they were all genuinely in the trenches — using AI for real marketing work.

The results split right down the middle.

First, the Good News: AI Really Has Gone Mainstream

More than half of marketers are already using AI for creative content, audience targeting, and customer service conversations. About half are using it for predictive analytics.

58% say they'll invest even more next year, especially in creative generation.

So AI in advertising isn't a "should we use it?" question anymore — it's a "if you don't, you'll be left behind" question.

But here's the thing.

Seven in Ten Have Already Watched It Backfire

The survey asked: have you experienced AI-related incidents in your advertising work?

70% said: yes.

What kind of incidents?

AI fabricating content — what's known as AI hallucination. The output looks plausible, but it's fake. Other outputs carried bias, or the content was inappropriate. Some drifted off-brand voice, or even offended users. Compliance failures kept cropping up too.

The consequences? 40% have had to pull an ad. More than a third have dealt with the PR fallout of brand reputation damage. Nearly 30% were forced into internal audits.

Only 6% said: the impact was minimal, no big deal.

Picture a team that spent months building a campaign. It goes live, then a single piece of AI-generated misinformation blows it up. The ad has to be pulled, the budget goes up in smoke, and the brand takes a public beating.

This isn't hypothetical anymore — it's already happening.

The Number That Surprised Me Most

The survey asked: do you feel prepared to handle AI risks?

Nearly 90% said: yes, we're prepared.

Then, in the same survey, they were asked whether they have formal AI governance tools. Things like automated content detection, bias audits, red team testing (deliberately attacking your own AI models to find the weak spots).

Only a third of brands, agencies, and publishers have them — or even plan to.

IAB's other report (State of Data 2025) confirms the same gap.

Do you see what's happening here?

The vast majority feel "we're fine" — but the actual protective tooling they own is vanishingly thin. 10% of respondents even admitted they're not doing any AI risk management at all, or they're not sure what they're actually doing.

This is a textbook false sense of security. It's like someone saying "I'm a great driver, I don't need a seatbelt" — but they've already scraped the guardrail seven times.

So What's Actually Keeping Things in Check?

Most teams' current AI risk management comes down to two things: manual review, and a brand compliance checklist.

Are those useful? Yes. Are they enough? Nowhere close.

When you're generating hundreds of ad creatives a month with AI, eyeballing each one and ticking boxes on a list simply can't keep up with that volume. The more advanced approaches? Red team testing, automated detection tools, bringing in outside AI ethics specialists. Very few are using them.

Plainly put, the industry's defenses are still stuck in the "artisan workshop" stage — while AI's output velocity has already gone industrial-scale.

The gap between those two is exactly where risk takes root and grows.

The Other Mess: Who's Actually in Charge?

The survey asked a very practical question: in your company, who owns AI governance?

The answers were all over the map.

Some said senior leadership handles it. Some said there's a dedicated AI task force. Some pointed to the marketing team, others to legal and compliance, others to data science or MarTech teams.

17% of companies brought in external partners to do it.

But 14% of companies have no one in charge. And another slice isn't sure who owns it at all.

What does "no one in charge" mean? When something goes wrong, nobody owns it. Risk just slips through the cracks.

But People Genuinely Want Help

There was another number in the survey that stuck with me.

The question: if a third-party solution existed that could detect AI hallucinations, bias, and off-brand drift for you — would you consider it?

Over 90% said: yes.

One person summed it up as wanting "peace of mind." Another said it would reduce risk to both brand and business.

Only a tiny minority felt it wasn't needed, either because they trusted their internal team or because they worried about cost.

The signal here is unmistakable: marketers know they can't handle this alone. They're waiting for professional tools and people to show up.

What IAB Wants to Push Forward

IAB closed the report with a four-step action plan. I read it and thought it was pretty solid — translated into plain English:

First, pin governance responsibility on specific individuals. Don't let AI risk become "everyone owns it, which means no one owns it." Senior leaders need to step in, and ideally a cross-functional task force gets formed.

Second, upgrade basic checks into system-level capability. Manual review and checklists are a starting point — but you have to climb higher. Structured assessment processes, automated audits (with humans kept in the loop at critical checkpoints), continuous monitoring that can scale with your content volume.

Third, don't tough it out alone when you could lean on outside help. Third-party tools and specialist teams exist precisely for this — they let you move fast without crashing.

Fourth, transparency. Don't just say "we use AI responsibly" — show the evidence. Build tracking systems that record how AI is used, where the risks are, and whether audit logs can be pulled up on demand.

At the End of the Day

The picture these numbers paint is clear: the advertising industry is running fast, but the guardrails haven't caught up.

AI's capabilities are genuinely exciting. Content production efficiency, audience targeting precision, interactive experiences — all of these have jumped a level compared to before. But seven in ten have already fallen into a pit, and the overwhelming majority still feel like they're fine.

I know this feeling too well. It's like the early days of a startup, when the growth numbers are so intoxicating that nobody wants to look at the pile of technical debt stacking up in the backend. Then one day the server crashes, and you realize the cost you saved back then now has to be repaid tenfold.

AI in advertising is repeating that story. The difference is, the debt this time isn't only technical — it's brand trust debt.

The good news is that over 90% are willing to open the door and accept help. That tells you they actually know what's going on underneath.

The remaining question is: who's going to actually push that door open and deliver the standards and the tools.

IAB is doing it. Let's hope more organizations follow.

---

Data source: IAB and Aymara, via the Insights Engine platform (powered by Attest), surveying 125 U.S. advertising industry executives from companies with 50+ employees who are directly involved in or aware of their company's use of AI in advertising and marketing.