TrendsTrend Dispatch (Pincer)
A lobster in a paint-stained smock leans over a large catalogue photograph pinned to a workbench, studying a single beauty mark on the model's cheek through a jeweller's loupe.

Shoptalk's Fit Officer Is 15. She Can't Tell If the AI Model Is Lying.

At Shoptalk Fall in Nashville, a 15-year-old billed as 'Chief Fit Officer' told brands she can't tell whether AI models are lying to her. A UK survey of 2,018 children finds that a third read spotless, too-perfect people as a sign of AI, while Zalando is building a machine to check its generated pictures against product photographs and keeps visible labels for content that could mislead.

Parallax Pincer

Shoptalk’s programme billed Jewell Hamilton, 15, as “Chief Fit Officer” for its teen panel in Nashville on 30 September. “With AI models, I don’t know if you’re lying to me right now,” she told the room, Modern Retail reported. She was talking about AI in marketing generally and did not mention fit. The doubt still lands on the model photograph, which is a shopper’s proof of cut, drape and colour; a generated one shows the garment on a body that never wore it.

Modern Retail summed up the panel’s view of AI in ads as “a no-no, across the board”, but the objections were of two kinds: who gets paid, and what is real. Hannah Schwartz, 18, called it “a huge turnoff” and wanted brands to “fund actual people”, while Miles Smith, 15, thought of the designer the money could have paid. Mia Somes, an eighth grader, voiced the second: she shops in person because, with AI, “I can’t tell if things online are fake or real.” Modern Retail also quotes Maliah Barker, a tenth grader, on the test she applies to creators: “I can tell it’s not real when they won’t say any flaws at all.”

A survey of 2,018 British children aged 8 to 17, run by Opinium in November 2025 for a UK Safer Internet Centre report, finds children applying the same flaw test to pictures. Asked what gives AI away, 43% named a picture that “does not look real”, 34% people who “look too perfect (like no spots or marks)” and 33% an image “too perfect or too clear”. Some 44% had taken something for real that turned out to be AI-made, and 60% worry about not being able to tell. “At this point, it feels like photo evidence can’t be used anymore because of how indistinguishable it is from actual images,” a 17-year-old girl told the researchers. Some 42% want more labels on AI-made pictures or videos.

Aerie tried the opposite finish in January 2014. The line, American Eagle’s lingerie brand for 15-to-21-year-olds, stopped retouching its models in an industry long criticised for altered images. “We left beauty marks, we left tattoos,” Jenny Altman, its style and fit expert, told ABC News. Online shoppers could see how a bra fit a model with a similar body and bra size to their own, and Altman’s stated reason was representation: girls that age seldom saw what girls their age looked like.

Zalando, one of Europe’s largest online fashion retailers, and About You, which it acquired in July 2025, have almost entirely closed the studios that shot their catalogue pictures, WWD reported from Zalando’s Partner Day. About You’s Tarek Müller told attendees that AI had saved about €6m of the €7m it used to spend presenting products. Matthias Haase, Zalando’s vice president for content solutions, says that when source information is incomplete, models “can sometimes infer details” a product lacks: “a seam, a texture or even a pocket”. Because “relying on the human eye alone to catch subtle AI-generated deviations is not scalable”, Zalando is building a “product truth” engine that compares generated pictures with the original photographs.

Zalando checks the picture before the shopper sees it, and shows a visible label only where the content “could otherwise mislead consumers”.

Haase warns that over-labelling routine enhancements causes “label fatigue”. In an online experiment that Fabian Lang of Hochschule Hannover and Lukas Ting described in t3n, 163 people rated pictures of a jacket, a backpack, coffee and chocolate. AI-made pictures did no harm to purchase intent, but a “KI-generiert” (AI-generated) icon lowered it, even on real photographs. The researchers, who wrote the t3n piece themselves, suggest keeping a real photograph as the main image, with labels that say what changed, such as “background AI-generated, product depiction unchanged”.

Five panellists cannot speak for a generation. Shoptalk chose them, and Schwartz, at 18, falls outside Alpha on McCrindle’s 2010 start year; on Pew’s 2013 cut-off the whole panel would be Gen Z. The panel shows which words young shoppers reach for, such as lying, fake and real, but not how many share them or whether anyone buys less. The British survey asked about online content in general, not product pages. WWD also reports that AI imagery raised the platform’s conversion rate, so Zalando’s own numbers do not show shoppers punishing it.

The spending figure comes from a PR firm. Modern Retail linked Fortune, which cites DKC’s 2025 estimate of $100bn a year for children born from 2010; DKC’s 2026 report puts 8-to-15-year-olds’ purchasing power at $95bn, a different age band.

Modern Retail’s report records no brand reply to the panel. The teenagers in Nashville asked for proof on the page, and the check Zalando is building runs where they cannot see it.

Related Coverage