Everyone is already using AI. The only question left is where you let it touch the audience.
Seventy-eight percent of the world’s largest advertisers now use AI-generated or AI-enhanced content in their marketing. So, let’s stop pretending there’s a decision to make about whether to use AI. There isn’t. The decision is where we let it touch the audience, and where we deliberately keep a person in front.
That’s a harder question than it looks, because our industry is currently misreading the room.
The 37-point blind spot
Ask ad executives how young consumers feel about AI-generated advertising and 82% say positive. Ask the consumers themselves, and it’s 45%.
That’s a 37-point gap between what we think our audience feels and what they actually feel. And it’s getting worse. The same gap was 32 points in 2024. We are confidently wrong about the people we’re paid to understand, and the confidence is growing faster than the accuracy.
What audiences actually punish
Here’s the part that surprised us, and it’s worth being honest about: the research doesn’t show that human-made content simply performs better. Audiences can’t reliably tell AI work from human work, and when they can’t tell, the AI work performs just as well.
What costs you is content that reads as generic, effortless, or slightly off. The moment people suspect they’re being handled by a machine, engagement drops. Not because a machine was involved, but because it showed.
So, this is a craft problem before it’s a technology problem. And the most reliable way to make work that doesn’t feel machine-made is, still, to not make it with a machine.
The “nobody will notice” era is over
Since 2 August, EU rules require brands to disclose AI-generated images, audio and video, with no grace period and fines reaching €15 million or 3% of worldwide turnover. Meanwhile, the content is starting to carry its own receipt: image and audio generators now sign their output with content credentials and invisible watermarks, some text tools watermark by default, and TikTok, Meta and Pinterest auto-label what they detect.
Look at the brands that pushed too far. Google pulling an Olympics ad, McDonald’s taking down a Christmas spot after three days, Klarna rehiring the customer service people it replaced. Nobody was fined. They were embarrassed, and it cost more attention than the work ever earned.
Where we use AI, and where we don’t
We’re not the anti-AI agency. AI is genuinely excellent at the work nobody was ever going to admire: resizing and localizing something a person already art-directed, generating twenty first drafts to react to, building backgrounds and product shots with no people in them, handling targeting, tagging, transcripts and reporting. Brands doing this well are cutting production time and cost by a third to a half. That’s real money, and we take it.
So, we ask one question before every brief:
Does the AI touch a person, a face, a voice, or a real experience?
If yes, we keep it human. Customer stories, real people and places, your team, anything presented as first-hand.
If no, we automate it freely and put the hours we save back into the work that needs a person.
The efficiency is real, and we intend to take all of it. We just don’t think it should be paid for with your audience’s trust.
Wondering where the line sits for your brand? That’s a conversation worth having before the next campaign, not after it.