Brand strategy
Research into luxury brands shows when AI actually damages your brand: the moment emotion is involved. Where that line lies, and what it costs.
Your client doesn't notice when AI writes your opening hours. They do when AI writes your thank-you email. That difference has a name: brand authenticity.
For years that sounded like a gut feeling from designers watching their profession change. Now there's a measurement, and the effect is larger than I expected.
AI-written communication only costs you credibility when emotion is involved. Factual texts and editorial work remain unpunished. The boundary doesn't lie with the technology, but with who determines the meaning. Below, where that boundary runs and what brand authenticity costs you if you ignore it.
Luxury is the laboratory of the brand world. That's where the gap between price and meaning is greatest, and that's where it's most sharply studied. Anyone selling a three-thousand-euro bag isn't selling leather.
The mechanism isn't reserved for the top of the market. As soon as your client chooses based on feeling rather than price, the same rules apply. A physiotherapy practice that sells trust, a caterer who sells an evening, a gym that sells the feeling that someone belongs: all brands that must live on meaning.
For you the question isn't whether you're luxury. The question is whether your client buys from you for what it is or for what it means. In the latter case, brand authenticity is your subject.
Colleen Kirk and Julian Givi published in 2025 in the Journal of Business Research a series of seven pre-registered experiments, together involving over two thousand participants. They had people read the same marketing communication, with one difference. One group was told a human had written it, the other that AI had done it.
For an emotional email, willingness to recommend the brand dropped from 5.45 to 4.11 on a seven-point scale. That's not noise. That's a brand losing part of its word-of-mouth, because there's a line beneath an email.
The researchers also found out why. It's not about quality, because the texts were identical. It's about brand authenticity, and for some readers about what the researchers call moral disgust. The reader feels manipulated. Emotion was feigned that wasn't there.
This is where it gets interesting, and where most articles about this subject go wrong by only reporting the shock reaction.
The effect disappears with factual content. A price list, a manual, directions: there the reader doesn't care who or what built the sentences.
The effect also disappears when AI only edits and the human remains the sender. And with repurposed content, a text that was already copied, it even reverses. Then people value the AI version more highly.
That's not a detail. That's the working instruction.
The boundary doesn't run between using AI or not. The boundary runs between tasks where meaning is produced and tasks where it isn't. As soon as you as a brand claim to feel something, that feeling must come from somewhere. As soon as you're only informing, the machine may do its work. That's where brand authenticity begins and ends.
That's the strongest objection, and it has a point. In the experiments it said so in so many words underneath. In practice nobody puts beneath a thank-you email that a model drafted it. The researchers say so themselves: they worked with invented situations and with participants from online panels in the United States and the United Kingdom. The measurement is therefore stricter than daily reality.
But the AI label isn't the cause. It's the revelation.
The text was already empty before anyone read it. The label underneath only made visible what wasn't in it. Remove it and the emptiness remains, only then the reader can't name it. They only feel that the email leaves them cold, and they don't forward it.
Besides, that label is increasingly becoming superfluous. I catch myself spotting it from a distance now: four sentences with the same cadence, a list of three where two would suffice, an enumeration going nowhere. What sounded smooth last year now sounds like a model trying its best. Anyone working with these systems daily recognises them within two paragraphs. That audience is growing, and your clients are in it. I wrote earlier about why humanity is your greatest asset in the AI era.
For Bonaire Eco Rentals, a family business that rents sustainable holiday homes on the island, three values had to come together in a brand identity: ecological responsibility, luxury comfort and nature conservation. On paper that's a tension, because luxury and austerity clash as soon as you don't take them seriously.
The three of us flew to the island. Not for atmosphere, but because the research was there. We photographed at sunrise, we saw the coral to which the company owes its raison d'être, and we came back with what later became the Palette of Bonaire: ruby-red coral and turquoise water, taken from the island itself and not from a colour chart.
Today I can have an image made in half a minute of turquoise water with a coral-red accent colour. It would look good. It would also be worthless, because it's not that water, not that light and not that house. The founder said at the time that we'd captured the complete essence of the business. That was only possible because there'd been a team that had seen that essence.
That's brand authenticity in an image. An AI model can imagine a sunrise. It can't attend one.
Valentina Chiaudano and Hafsa Shakil from the University of Turin analysed 79 academic articles about AI in luxury brand management, published up to and including October 2025. Their overview came out this month and delivers precisely the concept that's missing in our field.
Brand authenticity, they write, is no longer the absence of technology in the AI era. It's the degree to which technology aligns ethically and aesthetically with where the brand comes from. They call that "curated authenticity": authenticity as the outcome of a conscious directorial choice rather than as a natural state.
Additionally, they describe what they call "hybrid intelligence". Value emerges where prediction and intuition meet, where automation and craft work together rather than replacing each other.
A caveat about that second source. This is a literature review, not a measurement. The authors summarise what others found, the journal has only existed since 2024 and there's no limitations section in the article. Where this research and that of Kirk and Givi would contradict each other, the experiment wins. But they don't contradict each other. Kirk and Givi measure the boundary, Chiaudano and Shakil give it a name.
I work with AI in virtually every project, and I have a division of roles for it that runs precisely along that boundary.
I determine the brand, the strategy, the story and the judgement. That stays with me, because that's where meaning emerges. AI produces: variants, structures, first versions, analyses. For code and technology I use different tools again. And I decide and deliver, because my name is underneath. As long as I don't want or can't make that connection, the project isn't finished.
In practice, that means four things. I write a customer story myself and have it edited at most. A manual, a specification or a list I'm happy to have drafted and then check afterwards. Images that need to show a place that actually exists I photograph, and images that need to show an idea may be generated. And when in doubt I ask a question: does this text claim to feel something? If yes, then there must be a person behind it who felt it too.
What strikes me since reading this research: those weren't principles born of caution. That simply turns out to be where the effect stops.
It's worth measuring your own communications against this standard. Not to remove AI, but to see where you're accidentally letting it do the wrong work. That's precisely the work I do with business owners in a Brand Direction Programme.
Look at your welcome email, your thank-you email, your story about how the company began and your response to a complaint. Those are the places where your brand claims to feel something. Then look at your product copy, your frequently asked questions, your directions and your newsletter approach. The machine can safely help with those.
The mistake I see in the market isn't that business owners use AI. It's that they deploy it on exactly the wrong half, because that half is the hardest and the writing takes longest there. That's how brand authenticity leaks away at the only place where it counts.
Your brand isn't what you say you are. It's what's left when someone works out how it was made. Brand authenticity isn't an attitude, then—it's a division of labour.
If you'd like to know where that line runs in your communications, send me an email. No form, no funnel, no sales call. Just an email.
Kirk, C.P. and Givi, J. (2025). The AI-authorship effect: Understanding authenticity, moral disgust, and consumer responses to AI-generated marketing communications. Journal of Business Research, 186, 114984.
Chiaudano, V. and Shakil, H. (2025). Understanding the role of artificial intelligence in luxury brand management. Journal of Emerging Perspectives, 2, 33-44. Published online 18 July 2026.
Written by
Peter van der SteegePeter van der Steege is a brand strategist, designer and AI director. He builds brands for entrepreneurs and writes about what makes brands strong, from strategy to the role of AI and humanity. He lives and works in Groningen.
