Amazon now requires sellers to identify media containing photorealistic AI-generated people with specific metadata. The requirement is broader than product listing images: Amazon says it covers buyer-facing media including listing images, A+ Content, Brand Story, Store images, product videos, A+ videos, Store videos, and advertising videos.
Need help reviewing your Amazon content and marketplace compliance? Visit Alisterr.com.
What Amazon requires
For media containing a photorealistic person who was entirely generated by AI, Amazon requires the file to carry the keyword:
contains-synthetic-performer
Amazon specifies that the keyword must be added to the dc:subject field in the XMP metadata using an IPTC-compatible metadata editor before the media is uploaded. Amazon says it uses this metadata to add an indicator for customers where applicable.
This is not a general requirement to label every image created with AI.
The deciding factor is whether the media contains a photorealistic AI-generated person.
Which images need the tag?
| Media | Tag required? |
|---|---|
| Photorealistic person entirely generated by AI | Yes |
| Photorealistic AI-generated person shown only partially | Yes |
| Image contains no people | No |
| Real person whose image was edited with AI | No |
| Non-photorealistic person | No |
| Movie, TV, game or other expressive-work character | No |
| AI-generated background with no AI-generated person | No |
Amazon specifically says that media showing only real people does not require this metadata even when AI tools were used to alter the image. It also excludes non-photorealistic people, expressive-work characters, and media with no people.
For partial images, Amazon’s clarification is important. A cropped image does not automatically escape the requirement. If the visible portion still depicts a photorealistic AI-generated human, Amazon says the metadata should be applied.
The exact metadata sellers need to add
The requirement is not simply adding “AI-generated” to a file name or description.
Amazon specifies:
XMP field: dc:subject
Keyword: contains-synthetic-performer
The metadata should be embedded in the actual image or video file before upload using an IPTC-compatible metadata editor.
That distinction matters because adding a note to your Seller Central workflow without embedding the required metadata does not satisfy Amazon’s stated tagging process for non-A+ media.
How to tag files before uploading
The practical workflow is straightforward:
1. Identify media containing AI-generated people
Start with every buyer-facing asset, not just the main product images.
For each asset, determine whether the person is:
- entirely AI-generated and photorealistic
- a real person modified with AI
- non-photorealistic
- a fictional character from an expressive work
- not present at all
Only the first category falls within the metadata requirement described by Amazon.
2. Keep the original files
Create a separate working folder for the assets that need tagging.
Do not overwrite your only original copy. Keep an untouched backup and a separate tagged version so you can compare the files later.
3. Add the exact keyword
Using an IPTC-compatible metadata editor, add:
contains-synthetic-performer
to:
XMP → dc:subject
Amazon’s requirement is specific to this metadata field.
4. Verify the metadata before upload
Do not assume that clicking save means the metadata was successfully embedded.
Open the resulting file in a metadata inspection tool and confirm that the keyword appears under the XMP dc:subject field.
This gives you a pre-upload verification point and avoids discovering after upload that a batch was processed incorrectly.
5. Upload through the relevant Amazon workflow
For listing, Store, Brand Story, video and advertising media, Amazon says the file-level metadata must be present before upload.
A+ Content has an additional option.
A+ Content has a separate tagging option
Amazon says A+ Content supports two methods.
You can embed contains-synthetic-performer in the image’s XMP dc:subject field before uploading, or use the “AI-generated people” checkbox in A+ Content Manager’s Creative Assets workflow. Amazon says it will apply the metadata when that checkbox is selected.
Amazon also says images generated using its A+ generative AI features, such as Creative Studio, are tagged automatically.
That means the process is different depending on where the asset is being used.
For A+ Content, you have an Amazon-side tagging option.
For other buyer-facing media, Amazon’s clarification says the file needs to be tagged before upload.
What about older images already published?
This is one of the areas where Amazon’s public forum replies need to be read carefully.
In one Amazon forum response, Amazon said compliance is mandatory going forward for new uploads and described updating older images as a best practice when making listing edits.
In another detailed Amazon response, however, Amazon explicitly said that compliance has been mandatory since June 6, 2026, and instructed sellers to retroactively add the metadata to content published before that date. The same response says noncompliance could result in violations and potential ASIN removal.
Because Amazon has published this stricter retroactive instruction, sellers with older assets should not assume that previously uploaded media is automatically exempt.
The safer documented workflow is to audit older buyer-facing media and identify assets containing photorealistic AI-generated people.
How to handle a large catalog
Amazon’s forum guidance specifies the metadata that must be present, but the Amazon responses reviewed here do not document a Seller Central control that automatically applies this metadata to thousands of existing files in bulk.
For a large catalog, the bulk operation therefore needs to happen at the file-processing stage.
A practical workflow is:
Export or collect original media → identify qualifying assets → batch-embed the XMP keyword → verify metadata → upload the corrected files through the relevant Amazon workflow → keep an audit record.
For example, a catalog with hundreds of qualifying images should not be processed manually one file at a time if the metadata tool supports batch processing. The important point is that “bulk tagging” refers to processing multiple files with the required metadata, not to a documented Amazon bulk-tagging button.
How to handle A+ Content differently
A+ Content deserves its own audit because Amazon provides the additional AI-generated people checkbox.
For new A+ assets:
- Upload the asset in A+ Content Manager.
- Identify whether it contains a photorealistic AI-generated person.
- Use the AI-generated people option where applicable, or upload a file already containing the required XMP metadata.
- Confirm the asset is processed correctly before publishing.
For assets created through Amazon’s own supported A+ generative AI tools, Amazon says the tagging is automatic.
For existing A+ content, review older assets rather than assuming the original upload date makes them exempt.
What sellers should not do
There are several easy ways to over-apply the rule.
Do not tag every image simply because AI was used somewhere in the production process.
Amazon says metadata is not required for an image that only contains real people, even if those people were altered with AI. It is also not required for images with no people or people who are not photorealistic.
Do not treat an AI-generated background by itself as proof that the image needs the synthetic-performer tag.
The documented trigger is the presence of a photorealistic AI-generated person.
Do not replace the required keyword with a generic label such as “AI generated.”
Amazon specifies the exact keyword contains-synthetic-performer.
Why Amazon introduced the requirement
Amazon says the requirement is connected to legislation requiring disclosure when advertisements contain photorealistic AI-generated people.
New York, for example, enacted legislation requiring disclosure when advertisements use AI-generated synthetic performers. The law took effect on June 9, 2026.
Amazon’s marketplace requirement is broader than simply following one state’s advertising rule. Amazon says its metadata requirement applies across its worldwide stores and covers multiple types of buyer-facing media.
Amazon AI media tagging checklist
Before publishing an asset, confirm:
Person check: Is there a person in the image or video?
AI check: Was that person entirely generated by AI?
Photorealism check: Does the person appear photorealistic?
Metadata check: Does the file contain contains-synthetic-performer in XMP dc:subject?
Platform check: Is the asset being uploaded to a workflow that requires file-level metadata, or is it A+ Content where Amazon provides its own tagging option?
Verification check: Has the metadata been inspected before upload?
Legacy check: Have older buyer-facing assets containing qualifying AI-generated people been audited?
Final takeaway
Amazon’s requirement is narrower than “tag every AI image,” but broader than “tag only the main listing image.”
The documented trigger is a photorealistic AI-generated person. When that trigger applies, Amazon specifies the exact XMP metadata keyword:
contains-synthetic-performer
For A+ Content, sellers can use Amazon’s AI-generated people checkbox or embed the metadata themselves. For listing, Store, Brand Story, video and advertising media, Amazon’s current forum guidance points sellers toward tagging the file before upload.
For large catalogs, the key is to separate content classification from metadata processing. First determine which assets actually qualify. Then batch-process those files, verify the embedded metadata, and update the corresponding Amazon assets systematically.
That approach avoids both major mistakes: missing qualifying media and wasting time tagging assets Amazon says do not require the metadata.
Read more: Amazon Rufus Is Now Alexa for Shopping: What Sellers Need to Know About AI Shopping
Need help with Amazon marketplace content, compliance reviews, or catalog work? Visit Alisterr.com.
Leave a Reply