Amazon’s contains-synthetic-performer tag: what it is and how to add it (2026)

Amazon’s contains-synthetic-performer tag identifies images containing photorealistic people generated entirely by AI. Add that exact keyword to the image’s XMP dc:subject metadata before a new upload to product listings or A+ content; it is not a general label for every AI-edited product photo.

Which Amazon images need the synthetic-performer tag?

Amazon’s July 27, 2026 announcement covers photorealistic AI-generated people in product listing images and A+ content across its worldwide stores. The practical question is how the person was made, not whether the photo looks polished or whether the background was edited.

For example, a real shirt placed on a wholly generated, realistic model falls within the people-focused scope. A photographed bottle with a generated background but no people does not need this particular keyword under the announcement. That distinction does not settle other marketplace rules or another destination’s AI disclosure requirements.

Keep a record of the source photo and how the finished image was produced. A file’s appearance alone cannot reliably tell you whether its model was photographed or generated.

The final row comes from an official moderator clarification, rather than a guess that a hidden face removes the obligation. Do not decide using face visibility alone.

Does this replace Google’s AI-image metadata?

No. Amazon’s keyword describes synthetic people. Google Merchant Center’s AI-generated content guidance applies to all images created using generative AI and names IPTC DigitalSourceType TrainedAlgorithmicMedia.

A realistic generated model image may need both pieces of information. A generated product scene without people can fall outside Amazon’s synthetic-performer requirement while still needing Google’s AI-source metadata.

The fields are different. Writing the Amazon keyword into dc:subject does not populate DigitalSourceType, and writing a source type does not create the Amazon keyword. Review destinations separately using the marketplace image and AI rules guide.

Do older Amazon images need updating?

Use this working rule: tag new uploads, and update older relevant images when you edit listings. Do not describe old images as permanently exempt.

In one official reply, the moderator says new uploads must comply and recommends updating older images when editing those listings. Another moderator reply gives broader retroactive guidance and a different historical date.

Those replies do not justify publishing a single, certain deadline for every existing asset. For an older catalog, inventory the affected images, retain your original files, and check the current Seller Central instructions for the assets you are changing. This article does not turn a forum disagreement into a promise about enforcement.

Where does the keyword go inside the file?

It belongs in the XMP Dublin Core subject field, written as dc:subject. The exact keyword is contains-synthetic-performer, including the hyphens.

Renaming a file to include those words is not the same operation. Neither is adding them to a listing title, alt text or a folder name. Those locations do not write the requested field into the image.

Subject is a list that may already contain useful keywords. Inspect it before editing, and choose a method that preserves the information you need. A focused metadata change is preferable to deleting all metadata and trying to rebuild it later.

How do I add the tag with the free browser tool?

The AI image metadata tagger works locally with JPEG and PNG files. It can read existing AI tags in WebP, but it cannot write a tagged WebP download. Convert to a supported output format before the final tagging step.

1. Keep an untouched copy of your finished image.
2. Select the source-type option that accurately describes the image. The current choices include Made with generative AI, Edited with generative AI and No IPTC source tag.
3. Select Shows a photorealistic AI-generated person when that description is true.
4. Choose the JPEG or PNG files. Review the Tags now and After tagging columns.
5. Download the tagged image, or the ZIP when several images are ready.
6. Choose the downloaded file again and check that its existing-tags column reports the Amazon people tag.

The source-type selection is separate from the people checkbox. Do not select No IPTC source tag if you need to retain a Google AI-source classification, and do not assume every file in a mixed batch has the same requirements.

If the existing source classification is not among the radio options, do not choose an approximate replacement just to add Amazon’s keyword. Use the focused ExifTool command below instead.

There is an important limitation: this tool replaces existing XMP. It is not an archival metadata editor that merges every caption, rights field or subject keyword. Its writer keeps image data and other supported metadata, but unrelated information inside the replaced XMP packet is not retained. Use the ExifTool method below when you need an additive subject change.

How do I add it with ExifTool without replacing my keywords?

First, inspect a working copy. The official ExifTool XMP tables identify Subject as a list in the XMP-dc group.

exiftool -G1 -a -s -XMP-dc:Subject product-final.jpg

Then append the required value:

exiftool "-XMP-dc:Subject+=contains-synthetic-performer" product-final.jpg

The += matters. In ExifTool’s writing documentation, addition appends an element to a list; a plain assignment replaces its value. Appending preserves existing subject entries rather than replacing the entire subject list with the disclosure keyword.

Run this on files that need the keyword, then inspect the result. Check whether it is already present before repeating an append; do not build a routine that keeps adding the same keyword unnecessarily.

By default, ExifTool retains the earlier file with _original appended to its name. Keep that backup until you have verified the edited image. These examples deliberately omit overwrite-original options and broad deletion commands. A separate backup of your source folder is still sensible before batch work.

The same tag can be written to a supported PNG by using its actual filename. Do not change a filename extension to disguise the image’s real format.

When should I tag an image during editing?

Tag the final upload file after cropping, resizing, background work and format conversion. An earlier tagged master is useful, but it does not prove a later export contains the same metadata.

Treat each output as its own file. A square JPEG for a listing and a separately exported A+ asset can have different metadata even when they show the same scene.

CatalogCut’s image optimizer preserves recognized AI-disclosure fields in JPEG and PNG while removing private metadata. Its WebP handling removes metadata, so do not extend the JPEG/PNG promise to WebP. Other editing and delivery steps need their own check.

A small workflow change helps: keep an untagged working export, produce the tagged final, and label their folders clearly. Upload from the final folder after reading the tag back. This avoids choosing the visually identical file that came from an earlier editing step.

How do I verify the exact file I will upload?

Read both fields if the image will also be used in a Google product feed:

exiftool -G1 -a -s -XMP-dc:Subject -XMP-iptcExt:DigitalSourceType product-final.jpg

Confirm that the subject contains the complete Amazon keyword. If a source type is required for another destination, confirm that it is still present and appropriate. Also open the image itself to check that the correct product, variant and final crop were selected.

A local metadata read verifies the local file. It does not prove Amazon has detected the tag after upload or guarantee that a customer-facing indicator appears in every placement. The announcement describes an indicator where applicable; it does not promise a universal display or a seller-facing confirmation report.

What about the A+ content checkbox?

An official moderator describes an AI-generated people checkbox in A+ Content Manager Creative Assets as another way Amazon can apply the metadata during that upload workflow.

That is specific guidance about A+. Do not assume an option in one upload interface automatically tags a separate product-gallery file. Follow the current interface and instructions for the asset you are publishing, and retain your own tagged final when using the file elsewhere.

Are CatalogCut Stage downloads already tagged?

CatalogCut Stage PNG and JPEG downloads carry an IPTC AI-source tag. On-model downloads also carry Amazon’s synthetic-performer keyword, including sized exports and campaign ZIP images.

Stage’s workflow checks every generated image against the original before delivery and refunds credits for failed quality checks. Those checks and metadata are separate from your publication review. Compare the actual garment, labels, colour and construction before using the image.

If you edit the download afterward, verify the new export again. Metadata added to the original download cannot guarantee that another app preserves it.

Does tagging make an image acceptable to Amazon?

No. The keyword discloses a synthetic performer; it does not establish that the product is accurate, that a main image meets category rules, or that you have the necessary rights.

Use the marketplace photo checker for its supported file and framing checks, then separately review Amazon’s actual requirements. Neither a metadata editor nor a dimension check can approve a listing.

What to do next

Choose the exact JPEG or PNG you plan to upload and inspect it in the AI image metadata tagger. Keep the original, write only the disclosures that apply, and check the downloaded final before publishing.