AI product photography without misrepresenting your product: what is safe to make

AI product photography is safe when the product in every image is the product the buyer receives: only the setting is made, and the shape, colour, count, words and marks stay as photographed. There are two safe ways to do it, placing your real product's pixels into a made scene or drawing the scene with your product and checking every drawing against your photo, and the main image and any image that proves a fact should stay a photo. Walmart bars AI content that misrepresents a product, and Amazon asks for a tag on images with photorealistic AI-generated people.

AI tools can put your product on a marble counter, a beach towel or a sunlit desk in a minute, for a fraction of the cost of a shoot. They can also quietly change the product. A bottle comes back a little taller. A logo gains a letter. The ingredient list on the back turns into words that look right and say something else. A shopper who receives a different product from the one in the image has been misled, whatever tool made the image, and the marketplaces treat it that way.

This guide covers where AI product images go wrong, the two ways to make them safely, which image slots suit an AI scene, the marketplace rules that touch AI images, and a check to run before you publish.

Where do AI product images go wrong?

AI product images go wrong when the model changes the product. An image model does not copy your product into a scene: it draws a new image that looks like it, and the drawing can drift in ways that are easy to miss on a phone:

The last two rows carry the most risk. Invented words on a label make claims the real product does not make. A drawn mark can suggest a certification it does not have. Images make claims as much as words do: for health products, US regulators judge an ad by the whole impression it leaves, images and product name included.

Shoppers notice these gaps. Salsify's 2025 survey of US and UK shoppers reports that 54% had abandoned a purchase because of conflicting product information. An image that disagrees with the listing's words, or with the product in the box, is conflicting information.

What are the safe ways to make an AI product scene?

There are two safe ways to put a real product into an AI-made scene, and each has a trade-off.

1. Your real product, placed into a made scene. The scene is made first, with space left for the product, and your photo's pixels are placed into it. The product is never redrawn, so every word, colour and part stays exactly as photographed. The trade-off is realism at the edges: the light on the product comes from your photo, and the scene cannot wrap round the product in complex ways, such as a hand closing round a handle.

2. A scene drawn with your product in it, then checked. The model draws the whole image, product included, from your photo. Composition and light are more natural, and a hand can really hold the product. The trade-off is drift: shape, colour, count and especially fine print can change. This route is only safe with a check after every drawing, comparing it with your photo, and with the freedom to throw a drawing away.

A simple rule for choosing:

• Fine print matters (supplements, cosmetics, food, anything with ingredients, warnings or certification marks): use your real product placed into the scene.
• Shape and use matter more than small text (furniture, kitchen tools, clothing, bags): a drawn scene, checked, can work well.
• When unsure, choose the first. An unreadable label in a scene makes no claim. A rewritten label makes a false one.

Which listing images can be AI scenes, and which must stay photos?

Lifestyle, in-use and feature images can be AI scenes. The main image, the contents, the labels and anything else that proves a fact should stay a photo:

The pattern is simple. Anything that proves a fact stays a photo. Anything that shows a mood, a use or a setting can be made, as long as the product in it is still your product.

What do Amazon, Walmart and Etsy say about AI images?

Amazon, Walmart and Etsy each approach AI images differently:

• Amazon asks for a metadata tag on product images, videos and A+ content that contain photorealistic AI-generated people, from its notice of 22 July 2026. Its main image rules ask for the actual product, not an illustration. For supplements, it bars AI-generated label images. Our guide to Amazon's AI people tag covers the tag in detail.
• Walmart published guidelines on 3 April 2026: "All claims, images, and descriptions must match the actual product including size, color and other features", and "it's strictly prohibited to use AI-generated content that misrepresents a product." Sellers are responsible for reviewing and backing up AI content before it goes live.
• Etsy asks for original photos of the actual item buyers receive, not renderings or stock photos, and its Creativity Standards ask sellers to disclose AI use in the listing description.

Read together, the three marketplaces' rules are consistent: the product in any image must be the product the buyer receives. Only the setting is free.

How do you check an AI product image before you publish?

Put the AI image and your original photo side by side, both at full size, and check each line:

1. Shape. Trace the outline with your eye. Handles, lids, corners and straps in the same places, at the same proportions.
2. Colour. Compare in the same light on the same screen. No shift in shade.
3. Count. Count every piece. Nothing added, nothing missing, nothing detached that should be attached.
4. Words. Read every visible letter on the product. Each word should be on the real product, in the same place. Words that are blurred are acceptable; words that are different are not.
5. Marks and logos. Every certification mark and logo in the image is on the real product.
6. Size. A hand, a cup or a chair beside the product puts it at its real size.
7. People. For Amazon, note every image with a realistic AI-made person for the tag.
8. The setting. Nothing in the scene suggests an included accessory that is not in the box, or a result the product cannot deliver.

If an image fails any line, do not fix it by hand. Make another, or use the real-product route for that slot.

A second pair of eyes helps. The person who made the image has seen the product so often that they read what they expect to see. Someone who has not, holding the real product in one hand and the image on a screen, will spot a missing button or a changed word in seconds.

A worked example: a printed ceramic mug

Take a mug with a printed design and a care line on the base, sold on Amazon, Etsy and your Shopify store. A plan that keeps every image honest:

• Main image. Your front photo, cut out, on pure white. No AI.
• Second image. The mug on a kitchen counter in morning light, made with your real product placed into the scene, so the printed design is exactly as photographed.
• Third image. A hand holding the mug, drawn with the product in the scene for a natural grip, then checked: the handle in the same place, the design the same, no extra text.
• Fourth image. A size image: your photo beside its height and capacity, measured from the real mug.
• Fifth image. The base, photographed flat, with the care line readable.
• On Etsy. The drawn image of the hand stays off, because Etsy asks for photos of the real item, and the description says how the images were made.
• On Amazon. The drawn hand is part of an AI-made person, so that file carries the tag.

Five images, two made with AI, none of them showing a mug the buyer will not receive.

What should you ask an AI photo tool before using it?

Before you trust an AI photo tool with your listing images, ask:

• Does it redraw my product, or place my photo into the scene?
• If it redraws, does it check the drawing against my photo, and tell me what changed?
• Can I keep the words on an image separate from the image, so I can fix a typo without a new drawing?
• Does it know which marketplaces accept which images?
• Can I always get back to my original photo?

When should you use a plain product photo instead of AI?

A plain product photo is the better choice for products whose labels, finishes or one-of-a-kind making carry the sale:

• Labels that carry the sale. A supplement, a skincare serum or a food product with its claims on the front. Our page on supplement rules explains why the label photo matters.
• Exact finishes. Metallic, iridescent and transparent products are hard for any model to draw faithfully.
• Handmade and one-of-a-kind items. On Etsy especially, the real item is what the buyer is paying for.
• Anything you would struggle to defend. If you would not show the image to the marketplace with your photo beside it, do not publish it.

Your plain photos are also the source of every safe AI image. Our shot list covers what to photograph.

How Commicorn helps

Commicorn was built on one rule: we never invent the product.

• Two routes, your choice. Untouched product places your photo into a made scene with its pixels never redrawn. Best looking scene draws the product into the scene for more natural light and handling.
• Checks on every drawing. On Best looking scene, each drawing is checked against your photo. A drawing that fails is shown with the problem named, and it reaches your export only if you choose Use it anyway.
• Words never drawn. The words on a callout sit on a separate typed layer, never drawn into the image, and you can change them at any time.
• Each marketplace's rules on export. Images with an AI-drawn person carry Amazon's tag, and images where the AI drew your product stay out of your Etsy folder, because Etsy asks for photos of the real item.
• A main image from your photo. The Hero shot is your product on white, with only the light corrected.

Commicorn keeps your real product in every scene, or tells you when a drawing has changed it. Start free.

Questions sellers ask

Is AI product photography allowed on Amazon?

Yes, in the images after the main image. Amazon's main image rules ask for the actual product, not an illustration, on pure white, and since Amazon's notice of 22 July 2026, images with photorealistic AI-generated people need a metadata tag.

Does Etsy allow AI product images?

Etsy asks for original photos of the actual item buyers receive, not renderings or stock photos, and its Creativity Standards ask sellers to disclose AI use in the listing description.

Can AI change the words on my product's label?

Yes. A model that redraws the product can turn label text into letters that look plausible but say something else. Placing your real product's pixels into the scene keeps every word exactly as photographed.

Can supplement label images be AI-generated on Amazon?

No. As relayed by a labelling firm that read Amazon's help page, Amazon bars computer-drawn or AI-generated label images for supplements.

What does Walmart say about AI-generated product images?

Walmart's guidelines of 3 April 2026 say "All claims, images, and descriptions must match the actual product including size, color and other features", and Walmart strictly prohibits AI-generated content that misrepresents a product.