Amazon now requires sellers to add metadata to product images and A+ Content featuring photorealistic AI-generated people, so it can display a disclosure to shoppers when applicable. As more brands lean on AI to produce product content, this adds a new compliance step to the existing creative workflow, and puts metadata, not just the visuals themselves, at the center of listing management.
New AI Image Disclosure Requirements
The policy applies specifically to AI-generated people who appear as photorealistic individuals in images or videos, across product content and A+ Content worldwide. To comply, sellers need to use an IPTC-compatible metadata editor to add the keyword contains-synthetic-performer to the dc:subject (XMP) field before uploading eligible files. Amazon then displays a disclosure indicator to customers when applicable.
The requirement does not apply when images contain:
- Only real people, even if AI tools were used for editing
- Characters from movies, TV, streaming, documentaries, video games, or other expressive works
- No people at all
- People that are not photorealistic
Amazon points sellers to its tagging guidance along with the Product Image Guide and Shoppable Video Guide for implementation details. Whether content is produced in-house or through an agency, qualifying assets need to follow the metadata requirement before publication.
The Broader Impact of the Policy
This policy lands as generative AI becomes a bigger part of how sellers produce content. Amazon reported that Sponsored Brands campaigns using AI-generated images saw a 10.3% higher return on ad spend, brands using its AI creative tools advertised five times as many products, and advertisers using its Sponsored Brands image generator recorded 10% more sales per advertiser one month after adoption, based on internal U.S. data from 2024 and 2025.
Amazon-Reported AI Creative Results | Reported Outcome |
|---|---|
Sponsored Brands campaigns using AI-generated images | 10.3% higher return on ad spend |
Brands using Amazon AI creative tools | Advertised 5x more products |
Sponsored Brands image generator | 10% more sales per advertiser after one month |
Measurement period | Internal U.S. data, 2024 to 2025 |
Third-party sellers account for more than 60% of all units sold on Amazon, and AI-generated lifestyle imagery has become common because it is cheaper and faster than traditional photo shoots. That means disclosure requirements now touch content many sellers already rely on, particularly when it includes photorealistic AI-generated people.
The research on how disclosure labels affect shopper behavior is mixed. One study published in Equilibrium found AI-labeled ads generated lower consumer trust and purchase intent than human-made ads, especially for higher-priced products, while separate research in PNAS Nexus found AI labels reduced perceived accuracy and willingness to share content regardless of whether it was AI-generated or human-created.
This reflects a broader shift toward AI transparency across platforms. TikTok Shop and YouTube have introduced their own AI labeling measures, and additional U.S. state laws continue to expand disclosure requirements for AI-generated advertising and digital content.
Workflow Changes for Seller Compliance
Amazon recommends reviewing existing content, assigning responsibility across teams, and confirming required metadata stays attached before qualifying assets are uploaded. Key points for sellers:
- Audit existing product content
- Review new AI-generated assets
- Assign ownership across teams
- Verify metadata before upload
- Check that XMP metadata is preserved
Metadata preservation needs to become part of every content production process, since file exports and asset management systems can strip embedded information. A compliant master file can become noncompliant if that metadata is removed before publication, which makes verification an important final step for any asset containing AI-generated people.
Amazon frames this as a broader content management requirement built to support transparency in AI-generated advertising, not just a metadata tweak. Brands should know where qualifying content is being used and confirm the required disclosure stays attached throughout its production lifecycle.