Your customers can see the stitches that aren't there
J.Crew's AI catalog passed the glance and failed the people who know how a shirt is cut. The customers who look closest are the ones you lose.
In August 2025, someone scrolling J.Crew's Instagram saw a man in a striped rugby shirt and wanted it. The post was part of a campaign for a Vans collaboration, made to look like a page from the brand's late-'80s catalogs: boats, bicycles, an easel, that old-money ease. The commenter went to the website and came back empty-handed. "Can't find it in the site," they wrote.
J.Crew's account answered: "this is a vintage style."
The shirt had never existed. An image model made it up, along with the man wearing it.
Later that month the fashion newsletter Blackbird Spyplane reported that the images were AI-generated, which J.Crew hadn't said anywhere. A tip started it. What made the story was everything the newsletter and its readers found once they looked: a model's foot "bending backward underneath him at a 90-degree angle," shirt stripes "glitchy in the first slide, then crispy in the second," handlebars on a bike that looked wrong. One of the people they asked to look enlisted her father, who sails. The lifelines, he said, "should go all the way to the end of the bow. That gap left behind is where you don't want people falling through."
J.Crew's response was to edit the captions and credit the artist with the "digital art." They still didn't say AI.
The people who know the work
Most people probably glanced at those posts for a couple of seconds and kept scrolling, and I would have too. One of the newsletter's authors, a longtime fan of vintage J.Crew, admits doing the same at first. A generated image is built to pass that glance. It holds up for anyone who doesn't know how a shirt is cut, how a boat is rigged, or how a stitch sits.
Your best customers know exactly those things.
In April 2024, NBC News looked at crochet patterns on Etsy sold with AI-generated pictures of finished pieces nobody had ever made. Some of those shops had thousands of sales. The images, NBC wrote, "look great to the unassuming eye." The crocheters looked harder. Six reviewers posted photos of their half-finished projects, and the work didn't match the listing. A Reddit user with fifteen years of crocheting behind them described the moment they caught on: they "realized some of these were impossible and started looking closer for stitches that weren't there."
The image is fine until someone who knows the work goes looking for the stitches.
When the eye becomes a rule
In craft communities, that habit of looking closer is turning into a rule. This year the largest cross-stitch group on Reddit started asking members to post close-ups of their stitches, because, as the craft site CraftGossip put it, "one thing it's still hard for AI to replicate is individual stitches." A lot of people now show the backs of their pieces too.
The brands that get caught fastest are the ones whose customers make things themselves. In January 2024, Magic: The Gathering fans spotted a dial with "bizarre artifacting over the numbers" in a marketing image. Wizards of the Coast first said the art "was created by humans and not AI," then admitted otherwise and thanked its "diligent community." The same week, people flagged AI dragons in New Year marketing from Wacom, which makes drawing tablets for artists, and the artists who use them were among the loudest. Wacom said it had "vetted the images through a few popular online tools," and the tools had said they weren't AI. The people who flagged them didn't need a tool.
What the reaction looks like
Anger gets the attention, and some artists did tell their followers to buy from Wacom's competitors. The reaction I'd worry about more is the quieter one. Blackbird Spyplane called J.Crew's images "subtly but palpably wrong." One Etsy buyer wrote, "I will never purchase from this seller again." Both are a customer concluding the brand didn't look closely at its own product, and deciding to stop looking closely at the brand.
I think most marketing teams underestimate that cost. A generated lifestyle image saves the price of a shoot. What it spends is an assumption your most attentive customers make every time they see your work, that someone who knows the thing checked it.
Before it ships
I'd show the image to someone who makes the thing. The problem with J.Crew's boat took a sailor one look. Every company has, or can find, a person who knows how its product is stitched, poured, wired or cut. Their glance costs less than the shoot you skipped, and far less than the comments section.
If the image is generated, say so. The striped-rugby exchange is where J.Crew's campaign stopped being a style choice and became something closer to a lie: a customer asked a direct question about a product and got an answer that implied it existed. Disclosure has a cost. Being found out costs more, and it's paid by the customers you can least afford to lose.
The people who love what you make will always look closer than you did.
More on Craft
The more an AI avatar mimics a human, the more unsettling it becomes
Coca-Cola swapped people for polar bears and still landed in the uncanny valley. Let your AI look like AI.
Your perfectly mastered podcast now sounds like a bot
How AI audio mastering strips out the one signal that proves a voice is human
Cleaning up after AI is a growing job, priced like a quick fix
Freelancer.com listings to fix AI output rose 87% in ten months. The human arrives after every decision that mattered.