It was a Story. Of course it was a Story, the kind that vanishes in 24 hours, which is roughly the shelf life of most brand strategy these days. Adam Mosseri, the man who runs Instagram, stood in his weekly Q&A and answered the question every creative director in Los Angeles had been quietly hoping would go the other way.
Would Instagram punish AI content? No. “We do not throttle the reach of AI content just for being AI content,” he said, according to Social Media Today’s Oct. 19 write-up. The feed ranks what people like. If people like it, it travels. If it’s generated, it gets a label. That’s the plan.
Half the industry heard that as permission. We heard it as a warning.
Here’s why. On the very same day, Auros published a study conducted by Talker Research: 4,000 adults across the U.S., U.K. and Australia. In the U.S., 44% said they deliberately buy from brands that don’t use AI in their ads. Forty-three percent said they’d felt misled by a product advertised with AI. Only 20% of Americans could correctly identify real products that had been enhanced with AI.
Read those numbers together and the trap is obvious. People can’t reliably spot it. They don’t trust it. And they’re actively steering their wallets away from it. The algorithm is neutral. Your customer is not.
For a decade, brands outsourced their quality control to the feed. Bad creative got buried, so you could run ten mediocre variants and let the platform find the one that didn’t suck. The algorithm was the editor. The algorithm was the taste. You didn’t need a point of view; you needed a testing budget.
Mosseri just quit that job. Out loud. In a Story.
And let’s not pretend this was a neutral call. Meta has poured hundreds of billions of dollars into AI, and its apps now nudge you to generate images and video right inside the feed. Asking Instagram to down-rank AI content is asking a company to bury the product it’s selling you. That was never going to happen. The referee owns a stake in one of the teams.
So stop waiting for the platform to protect your brand from cheap work. It has every financial reason to encourage more of it.
Even Mosseri seems to know the bill is coming. The whole exchange started because an artist complained that AI accounts were copying his work, and Mosseri wrote on Threads that he’s very worried about where this is heading. His proposed fixes are labels and better tools for creators to fight copies. Useful, maybe. But neither one makes your content any good.
Labels are the real plan. Instagram intends to add more of them and wants creators to tag their own generated posts. So the platform won’t bury your synthetic campaign. It will do something worse. It will serve it to millions of people with a little badge on it, right in front of the 44% who’ve already decided what that badge means.
Think about what a label actually does in a feed. It’s not a ranking penalty. It’s a trust penalty, applied by the human holding the phone, in the half second before they decide whether you’re worth their thumb. A down-ranking at least hides your mistake. A label publishes it.
This is the part the “AI is just a tool” crowd keeps skipping. Nobody serious is arguing the tool is evil. The Auros data is weirdly clear on that: 53% of Americans say AI has helped them while shopping. They’ll happily let a chatbot find the lowest price on a rice cooker. What they won’t do is let a brand use the same machine to persuade them. Auros VP of Marketing Bobby Meixner put it plainly: consumers are skeptical when AI is used to persuade them.
So the distinction customers are drawing isn’t human versus machine. It’s help versus manipulation. AI as my assistant, fine. AI as your shortcut, I’m out.
And Harvard Business Review published research in September with a headline that should be taped to every in-house team’s monitor: AI-generated ads perform worse than human-made ones, even when customers can’t tell them apart. Sit with that. They can’t tell. It still doesn’t work as well.
That’s the most damning sentence in marketing this year, because it kills the last defense. The defense was always, “If nobody can tell, who cares?” Turns out the work tells on itself. Something in the frame is a little too clean, a little too average, a little too much like the median of everything the model ever saw. The customer may not be able to name it. Their behavior names it for them.
Now, the fair pushback. Some of you are thinking this is just an agency protecting its day rate. That we’d say all this whether the data existed or not. That every new production technology got the same sermon, from Photoshop to the iPhone to the ring light, and the world kept spinning.
Fair. And partly true. Every tool did get the sermon, and the people who refused to learn the tools got left behind. We use AI every day. For research, for rough boards, for variants, for the grunt work nobody misses.
But here’s what’s different this time. Photoshop didn’t make every brand’s ad look like every other brand’s ad. The iPhone didn’t produce a single statistical average of every photo ever taken. Generative tools do exactly that by design. They are averaging machines. And average was always the thing the feed punished.
Except now the feed has stopped punishing it.
Which means the only thing left between your brand and a sea of perfectly lit, perfectly forgettable, clearly labeled content is a human being with taste and the authority to say “no, that’s not us.” That person used to be a nice-to-have. A luxury for brands with real budgets. Now they’re the only quality control you’ve got.
Walk down Abbot Kinney on a Saturday and you can feel this without reading a single study. The shops with lines out the door didn’t win on volume. They won on a specific, slightly weird, entirely unmistakable point of view. The handwritten chalkboard. The playlist somebody actually chose. The barista who argues with you about natural wine. Nobody generated that. Somebody decided it.
That’s what taste is. Not good-looking content. Decisions. A thousand small decisions about what to leave out, what to keep rough, which take has the life in it even though the lighting is off. You can’t prompt your way to a point of view. You can only have one, or rent one, or fake one. And the faking is exactly what 44% of Americans are now voting against.
It’s why, when we built FAV 5 for Laced SouthBay, the whole series ran on real people, real picks and the friction of an actual conversation. No model can hand you that. The best moments in any shoot are the ones nobody scripted, and you only get them by being in the room and being smart enough to keep them.
So here’s where we land, and it’s not comfortable for anyone. Platforms have stopped being your editor. Mosseri said so. Your customers have started being your editor. Auros measured it. And the work that doesn’t come from a real point of view underperforms even when nobody can spot the seams. HBR published it.
Put those three together and the job description of a brand changes. You are no longer competing for reach. Reach is free now; Instagram will give it to anything people tap. You’re competing for the half second after reach, when a skeptical human sees a label, or doesn’t, and decides whether to trust you.
That half second is made of taste. Of craft. Of a decision someone was brave enough to make.
The feed won’t save you from your own slop anymore. It will just put a badge on it and hand it to the people most likely to walk away.
Make something nobody would ever mistake for the average.
