Every few weeks another brand pulls an ad. The pattern is always the same. Something goes out that was clearly machine made, the comments turn, it comes down inside 72 hours, and a thousand marketing people write posts about the death of AI creative.
Then the same marketing people go back to using AI on client work, because of course they do.
The number that should bother you
The IAB found that 82% of ad executives believed Gen Z felt good about AI-generated ads, while only 45% of Gen Z actually did (IAB, 2026). That is a 37-point gap between what the industry assumes and what the audience feels.
Separately, research using eye tracking and implicit response testing found that even polished AI-generated ads created a degree of dissonance in viewers and were less memorable than conventional advertising (NielsenIQ). Less memorable is the part that should sting. Not offensive. Not controversial. Just forgettable.
The three fixes everyone tried
The industry landed on a tidy consensus pretty fast, and all of it is beside the point.
Hide it. Use the tools, do not mention it, hope nobody notices. This works right up until somebody notices, and then you are not a brand that used AI, you are a brand that hid something.
Disclose it. Label the ad. Turns out labeling content as AI-generated makes people see it as less natural and less useful, so you get honesty points and worse performance.
Prove humans made it. Brands are now hiring people specifically to shoot behind the scenes footage showing that their ads were made by humans (Marketplace, 2026). We have reached the point where "we made this ourselves" is a selling point, like organic or fair trade.
All three treat authenticity as a question about production method. It never was.
What is actually going on
Go look at the ads that got pulled. Not the technique, the idea. Ask what the ad was actually saying.
In most cases the honest answer is nothing. There was no argument, no point of view, no specific thing the brand believed. There was a mood, some product, and a lot of surface. The AI did not cause that. The AI made it cheap enough to ship without anyone in the room noticing the idea was missing.
That is the real mechanism. Making things used to be expensive, and expense forced a conversation. Somebody had to justify the shoot. Somebody had to defend the concept to a person holding a budget. That friction killed a lot of empty ideas before they ever got made.
Remove the friction and the empty ideas ship. Then the audience does what audiences have always done to empty work, except now they have a name to call it.
A disclosure label cannot supply meaning the brand never defined.
Where the money went instead
The interesting part is that the backlash has been good for anyone who can still do the work.
A clothing brand ran an explicitly anti-AI campaign in early 2026 and the pledge tracked to a 23 percent sales lift on the prior quarter (Multiply, 2026). Another committed years ago to never replacing real people in its advertising and has spent the whole cycle being held up as the example. There is a real premium sitting there for brands with something specific to say and the willingness to say it in their own voice.
That premium is not available to you by announcing you are human. It is available by having a point of view that could not have come from anywhere else.
So how do we actually use it
We use AI constantly. Storyboards, variant testing, rough cuts, transcription, research, first-pass copy that gets torn apart. It is infrastructure. It is not the idea.
The rule we hold to is simple and a little annoying to enforce. The tools can accelerate anything that comes after the decision. They do not get to make the decision.
What the decision looks like in practice:
- What do we believe that our competitors would not say out loud.
- What is the specific moment, person, or detail this is built around.
- Would this still be recognizably ours with the logo removed.
If you cannot answer those three, the production method is not your problem. Making it faster only gets you to the wrong place sooner.
The short version
Audiences are not running a purity test on your workflow. They are reacting to whether there is anything in the work worth their attention. There usually is not, and that has been true for a long time. AI just removed the last cost barrier that was quietly filtering it out.
Have something to say. Then use whatever you want to say it.