It’s 6:40 on a Thursday and the client review is going well. Too well. The client loves the hero image, loves the line, loves the little animated bit at the end. Then someone from their side leans back and asks the question nobody prepped for: “Wait, did you use AI on any of this?”
The room does a thing. You know the thing…
If you’ve been in that room, you’re in good company. In a survey of 1,780 creative professionals, Envato found that 58% have used AI in client work without telling the client. Only 31% say they always disclose. And the group least likely to come clean? Agency owners, at 28%. The people who sign the contracts are the ones keeping the secret.
We’re not going to clutch our pearls about it. Most of that undisclosed AI is boring: a background extended, a transcript cleaned, a rough comp mocked up at midnight. Envato’s own data says 48% of the non-disclosers didn’t mention it because they don’t think every tool needs a footnote. Fair. Nobody discloses spellcheck.
But the silence tells you something important about the industry’s AI posture. There isn’t one. There’s a vibe. Some shops brag about going all in. Some promise they never touch it. Most quietly use it wherever it’s convenient and hope nobody asks at 6:40 on a Thursday.
All three are cop-outs. “All in” is a press release. “Never” is either bullshit or a slow death. “Wherever it’s convenient” is how you end up in that room with no answer.
The mature move is smaller and much less sexy. You write down where AI isn’t allowed to go.
A national razor brand just showed what that looks like in practice, and it’s worth studying precisely because it isn’t flashy. Dollar Shave Club launched “Perfectly Good Is Perfect” on October 20, a campaign across connected TV, YouTube, Meta and TikTok that pokes fun at the exhausting, expensive theater of modern self-care. The spokesperson sits in an ice bath with a Shake Weight, a green smoothie and an infrared mask before getting to the razors. Real people on camera. A “Red-ish Carpet” stunt that gave regular folks celebrity-style interviews in places like Cincinnati and Knoxville.
Where did AI go? Into the gags. When the spokesperson holds up six fingers to explain the six-blade razor, that’s AI. It also helped compress production to about two and a half months, according to Marketing Dive.
Where didn’t it go? The concept. And the shave itself, because, per the brand, the technology still can’t render shaving accurately. Laura Higgins, the brand’s chief brand and innovation officer, who wrote the brief, put the whole philosophy in one sentence: “We’ve had to become more choiceful around where we’re using [AI].”
Choiceful. Not “AI-first.” Not “AI-free.” Choiceful.
That’s an AI policy. It’s not a manifesto and it’s not a vibe. It’s a short list of places the machine is welcome and a shorter, more important list of places it isn’t. The idea is human. The product truth is human. The face of the brand is a human. The jokes that would cost a fortune to shoot practically? Have at it, robot.
Here’s why the no-go list matters more than the go list. Anyone can find places to use AI. Every tool on the market is designed to help you find more. The hard, valuable part of a policy is the boundary, because the boundary is where your brand lives.
And the boundary isn’t just a matter of taste anymore. It’s a matter of results.
In September, Harvard Business Review published research from Adam Peruta of Syracuse University’s Newhouse School comparing AI-generated and human-made video ads from major brands. Viewers mostly couldn’t tell which was which. The AI ads still came out weaker on both short-term sales potential and long-term brand equity. Think about how nasty that is. The work passes the eye test, sails through internal approval, and then quietly underdelivers in market. Nobody gets fired for it, because nobody can see it. The brand just gets a little softer every quarter.
Then there’s the audience. A Talker Research survey for Auros, released October 19, found that 44% of Americans go out of their way to buy from brands that don’t use AI in their ads, and 53% don’t trust AI-advertised products to be described accurately. Nearly half your customers are actively screening for the thing your team is quietly doing.
So no, “we’ll just use it wherever it’s convenient” isn’t a neutral position. It’s a bet that your customers won’t notice and that the research is wrong. Both bets look worse every month.
Here’s where it gets uncomfortable for us, and probably for you.
The obvious objection to all of this is that the no-go list is a luxury. Big brands with in-house teams and chief innovation officers can afford to be “choiceful.” Everyone else is drowning in content demands with half the budget and a third of the people, and AI is the only reason the calendar gets filled at all. Telling a stretched team to draw lines around its most useful tool sounds like advice from someone who has never had to post six times a week.
We hear that. We’ve lived it. And it’s exactly backwards.
The smaller you are, the more your brand depends on the things AI does worst. A local studio, a neighborhood restaurant, a med spa on Montana Avenue: their entire advantage over the big guys is specificity. The owner’s face. The real room. The regulars. The way the light actually hits the bar at 5 p.m. A big brand can survive a little genericness. A small one disappears into it. The no-go list isn’t a luxury for the little guy. It’s the moat.
There’s another reason to write it down now, and it has a date on it. On October 30, OpenAI will switch on “creative enhancements” by default for ChatGPT Ads, rewriting headlines and descriptions to fit the conversation, sometimes in real time, without per-version approval, according to Search Engine Roundtable. The platforms are not going to wait for you to develop a philosophy. Their defaults will become your policy unless you have one first.
So what goes on the list? It’s different for every brand, which is the point, but here’s how we’d frame the question. What is the one thing a customer believes about you that would feel like a betrayal if they learned a machine made it? That’s on the no-go list. For a restaurant, it’s probably the food. For a fitness studio, the trainers and the actual bodies in the room. For a professional services firm, the partner’s voice on LinkedIn. For a beauty brand, before-and-afters, full stop. Then ask: where does the machine save us time without touching that belief? Resizing. Rough comps. Subtitles. The six-finger joke. Go nuts.
Write it on one page. Share it with your clients or your agency. Tell your team. Then you’ll never have to do the thing in the room at 6:40 on a Thursday again, because the answer will already be written down and you’ll be proud of it.
Here’s the part nobody likes to admit. We’d bet most of the 58% didn’t stay quiet because they were ashamed of the tool. They stayed quiet because they couldn’t explain where they drew the line, and that’s because they never drew one.
A policy isn’t the list of what AI can do for you.
It’s the list of what you refuse to let it touch.
