Amazon now requires sellers to disclose when product images contain photorealistic AI-generated people — and it's telling shoppers directly. Not in a policy document buried in Seller Central. On the listing itself.
The rule stems from New York's synthetic performer advertising law, which took effect June 9, 2026. Amazon responded with a platform-wide mechanism: any product image or A+ Content featuring a photorealistic AI-generated person must have the keyword contains-synthetic-performer embedded in the image file's XMP metadata before upload. Amazon then shows a disclosure indicator to shoppers browsing that listing.
If your brand shifted to AI-generated model photography in the last 18 months — which many DTC brands did, given what studio shoots cost — this is affecting your live catalog right now.
What actually needs a tag
The rule is narrow but significant. It covers photorealistic AI-generated people only. Real models (including real photos with AI-applied color grading or background replacement), cartoons, illustrated figures, and product-only images are all exempt.
A lifestyle shot from a real studio with heavy AI post-processing? Not affected. A bedroom scene generated entirely in Midjourney or a similar tool, complete with a human figure? Affected.
The exemption for AI-edited real photos matters more than it first appears. If your workflow uses real models and you apply AI for lighting, skin smoothing, or background swap, you're likely clear. If your workflow generates the human figure itself — increasingly common for size-range representation and international market localization — you're in scope.
A+ Content is also in scope, which catches more brands than the listing image rule alone. Many DTC sellers use AI-generated lifestyle scenes in A+ without putting them in the primary product photo, precisely because the cost savings were easiest to justify there. Those images are not exempt.
How to add the tag
The contains-synthetic-performer keyword must be added to the dc:subject field in the image file's XMP metadata before upload. This is a file-level change, not a setting inside Seller Central.
Standard IPTC-compatible editors — Adobe Bridge, Lightroom, and most professional metadata tools — can write to XMP fields. The edit itself is not complex, but it requires touching each affected image individually before re-uploading. Amazon's seller interface has no bulk tool for flagging existing uploads.
Sellers who upload via API or third-party catalog management software need to verify that those pipelines preserve XMP metadata through the upload process. The tag must survive intact or the disclosure won't register correctly.
What shoppers see
Amazon has confirmed it will add a shopper-facing indicator to listings carrying the disclosure tag. Amazon has not published details on label placement or exact wording.
Research on AI content disclosures in other contexts generally shows lower engagement compared to unlabeled content. Whether that pattern holds for product photography specifically — where shoppers evaluate fit, scale, and material rather than image authenticity — is an open question. No Amazon-specific conversion data has been published.
But the question is no longer hypothetical. If you're running Sponsored Products or Sponsored Brands campaigns to listings where the primary product image now carries an AI disclosure, you're sending paid traffic to a page that looks different than it did. The assumption that AI-generated model imagery performs equivalently to real-model photography is being tested publicly, in real auctions, with real money.
The compliance exposure
New York's law puts liability on the advertiser, not the platform. Amazon is explicitly exempt as a distributor. Under the law, first violations carry a $1,000 civil penalty; subsequent violations are $5,000 each.
This is one state's law today. The EU AI Act's Article 50, which went into force August 2, requires disclosure for AI-generated ad content across EU markets. California, Illinois, and other states have introduced similar legislation. The regulatory direction is toward expanding disclosure requirements, not narrowing them. Brands building this into their creative workflow now won't be scrambling each time another jurisdiction follows.
What to audit this week
Start with your top 50 ASINs by Sponsored Products spend. Pull the image set for each one and identify any that use AI-generated human figures. Check A+ Content separately — that's where AI lifestyle imagery tends to accumulate without making it into the primary listing image.
For each affected image, the fix is at the metadata level: add the tag, re-export, re-upload. That's the compliance part. The harder strategic question: if your AI-model images deliver lower conversion after disclosure, what does real photography cost versus what you'd lose on conversion rate?
The free audit at Gromerce shows you which ASINs carry the most paid traffic — a useful starting point for prioritizing which listings to review first.
The compliance task is simple. The conversion question takes longer to answer, and that's the one worth starting now.
Sources: Search Engine Land, Amazon Seller Central, July–August 2026
