Effective July 27, 2026, Amazon requires sellers to add a specific metadata tag — the keyword contains-synthetic-performer in the XMP dc:subject field — to every product image and A+ Content video that features a photorealistic AI-generated person before uploading it to any listing, and sellers who have already uploaded AI lifestyle imagery without this tag are now in violation of Amazon’s product image policy and potentially in violation of New York General Business Law Section 396-b, which carried civil penalties of $1,000 for a first violation and $5,000 for each subsequent violation before Amazon’s requirement was even announced. The timing is not coincidental.
New York’s synthetic-performer advertising disclosure law took effect June 9, 2026, establishing a legal obligation for businesses that produce commercial advertisements containing AI-generated human likenesses to make a conspicuous disclosure. Amazon notified sellers on July 22 and made the requirement effective July 27, as MLQ News confirmed, — and the California AI Transparency Act became operative August 2, five days ago, adding a second state-level framework to the disclosure landscape. For sellers using Amazon AI generated images in lifestyle imagery, which has become standard practice as image generation tools have made photorealistic human lifestyle shots available at a fraction of traditional photography costs, this requirement is not a future consideration — it is a current compliance obligation on images that are live on listings right now.
This guide covers exactly what Amazon’s AI generated image requirement applies to, what it does not apply to, how to add the metadata correctly, which tools support it, what happens when Amazon finds non-compliant images, and the legal exposure sellers face under both New York and California law when AI-generated people appear in product imagery without the required disclosure.
What sellers need to do right now
If you are using AI-generated lifestyle imagery on Amazon listings — whether uploaded before or after July 27 — this guide provides the specific compliance steps, the legal exposure framework, and the enforcement risk assessment needed to act correctly. The requirement is narrow, technically specific, and addressable in hours for most sellers who understand exactly what it covers.
Quick summary: Amazon’s July 27, 2026 requirement applies to product images and A+ Content videos that contain photorealistic AI-generated people — not AI-edited photographs of real people, not AI-generated products without people, and not non-photorealistic illustrations. For qualifying images, sellers must add the exact keyword contains-synthetic-performer to the dc:subject field in XMP metadata using an IPTC-compatible editor before uploading. Amazon will display a disclosure indicator to buyers where applicable — and has stated that sellers are not required to add metadata retroactively to already-published content — but the underlying New York law has no such retroactivity exemption for content published after June 9, 2026, which is the date the statute took effect.
⚠️ Using AI lifestyle imagery on your listings right now? If those images were uploaded after June 9, 2026 and contain photorealistic AI-generated people without the contains-synthetic-performer metadata, you may be simultaneously non-compliant with Amazon’s image policy and in violation of New York General Business Law Section 396-b. Contact DAM Law Firm if listing enforcement from this requirement is affecting your account.
Table of Contents
- What the Requirement Applies To — and What It Does Not
- Why This Requirement Exists: The New York and California Laws
- How to Add the Metadata Correctly
- Which Tools Support XMP Metadata Tagging
- The Retroactivity Question: Existing Uploaded Images
- Building Compliance Into Your Image Workflow
- What Amazon Does When It Finds Non-Compliant Images
- The Legal Exposure Under New York and California Law
- When Non-Compliance Creates Listing and Account Enforcement Risk
- Frequently Asked Questions
- How DAM Law Firm Can Help
What the Requirement Applies To — and What It Does Not
The most important preliminary step is correctly identifying which images and videos in a seller’s catalog trigger the Amazon AI generated image metadata requirement. Amazon’s product image guidance is specific about the scope, and many sellers are either applying the requirement too broadly or missing the assets it actually covers.
Images and content the requirement covers
The Amazon AI generated image requirement covers product images and A+ Content videos that contain photorealistic AI-generated people — human figures that were entirely created by a generative AI model or software algorithm and that appear as realistic human beings in the image. A lifestyle shot showing a person using the product — where that person is a photorealistic AI-generated human rather than a model photographed in a studio — is the paradigm case, and a brand story video in A+ Content featuring an AI-generated human spokesperson also falls within the requirement. The common thread is that a human figure was generated entirely by AI and appears as a photorealistic person in the image or video.
Images the requirement does not cover
Several categories of AI-involved imagery fall outside the requirement. A photograph of a real human model edited or enhanced using AI tools — background removal, lighting adjustment, skin retouching — does not trigger the requirement because the person is real, not AI-generated. Products depicted in fully AI-generated images with no human figures are also outside the requirement regardless of how the image was generated. Non-photorealistic illustrations, cartoon figures, or stylized characters that are clearly not intended to depict real human beings are outside the scope. AI-generated images of hands, body parts without distinguishable faces, or heavily stylized human forms that do not qualify as photorealistic depictions of people are in a gray zone — Amazon’s guidance focuses on the “photorealistic” standard, and the safest interpretation for borderline cases is to add the tag rather than leave it off.
Where the content must appear to trigger the requirement
The requirement applies to product images uploaded to listing pages and to A+ Content images and videos. It does not, as currently written, apply to Sponsored Products or Sponsored Brands advertising creative — those are covered by separate Amazon advertising policies. Backend images in the image gallery that are not currently the main image or secondary images visible to buyers are still covered by the requirement when they are uploaded, because they remain part of the listing’s image set.
Why This Requirement Exists: The New York and California Laws
Amazon’s metadata requirement did not emerge from an internal AI content policy decision. It is a direct response to state-level legislation that creates legal obligations for businesses using AI-generated human likenesses in commercial advertising.
New York General Business Law Section 396-b
New York’s synthetic-performer advertising disclosure law — General Business Law Section 396-b — took effect June 9, 2026. The statute requires any business that produces or creates a commercial advertisement and has actual knowledge that it contains a synthetic performer to make a conspicuous disclosure. New York defines a synthetic performer as a human-like digital asset, created by generative AI or a software algorithm, that engages in a visual or audiovisual performance. Civil penalties run $1,000 for a first violation and $5,000 for each subsequent violation. The statute explicitly exempts platforms that merely publish or distribute commercial advertisements — which is why Amazon, as the platform, is not itself liable under the statute for seller-uploaded content, but sellers who create and upload the content bear the disclosure obligation directly.
California AI Transparency Act (SB 942, as amended by AB 853)
California’s AI Transparency Act became operative August 2, 2026 — five days ago. Its scope differs from New York’s statute in a way sellers must understand correctly: the California law places disclosure obligations on providers of generative AI systems with more than one million monthly users — the image generation tools themselves — rather than directly on sellers who use those tools to create content. However, the practical effect for Amazon sellers is that the AI image generation platforms being used to create lifestyle imagery are now subject to California disclosure requirements, which may change how those tools handle and label output. Sellers operating in California and using AI-generated human likenesses in commercial advertising should also evaluate their obligations under California’s broader AI and advertising disclosure framework independently of the Amazon policy requirement.
Why Amazon acted when it did
Amazon notified sellers on July 22 — six weeks after New York’s law took effect — and set compliance as effective July 27. The delay between the June 9 law effective date and Amazon’s July 22 notification created a window in which sellers uploading AI lifestyle imagery after June 9 may have been in violation of New York law without any warning from Amazon. Sellers who uploaded photorealistic AI-generated people in product images between June 9 and July 27 without any disclosure mechanism in place were operating under a legal obligation they may not have been aware of. That six-week window is the source of the retroactivity ambiguity discussed in Section 5.
How to Add the Metadata Correctly
Compliance is a technical step that requires adding a specific keyword to a specific field in the image file’s metadata before upload. Understanding each element of the requirement prevents the most common implementation errors.
The exact keyword and field
The keyword must be exactly: contains-synthetic-performer
It must be added to the dc:subject field in the XMP metadata of the image file. XMP (Extensible Metadata Platform) is a standardized metadata format embedded in image files, separate from the EXIF metadata that records camera and exposure data. The dc:subject field is the Dublin Core subject field — a standard metadata field used for tagging keywords and subjects. Spelling, hyphenation, and capitalization matter: contains-synthetic-performer in all lowercase with hyphens between the three words. Amazon will read this specific string to identify compliant images. A variation in spelling, case, or format will not trigger Amazon’s disclosure indicator and may not satisfy the underlying legal obligation.
The metadata must survive the upload pipeline
A critical technical consideration that most guides do not address is metadata stripping. Many image processing steps — file format conversion, compression, resizing, and export from design software — strip XMP metadata from the file, leaving only the visual content.
A seller who adds the contains-synthetic-performer keyword in a metadata editor, then exports the image through a pipeline that includes format conversion or compression, may deliver a file to Amazon with no metadata present even though the tag was added correctly at an earlier step. Testing the export pipeline by adding the tag to a test file, running it through the complete processing workflow, and verifying the metadata in the final file before upload is the only reliable way to confirm compliance survived the pipeline. Tools like ExifTool (command line) can verify the tag’s presence with the command exiftool -G1 imagefile.jpg and a search for the dc:subject field in the output.
Which Tools Support XMP Metadata Tagging
Sellers need an IPTC-compatible metadata editor to add XMP keywords. Several tools at different price points and technical levels can accomplish this.
Adobe Photoshop
Adobe Photoshop supports XMP metadata through File, then File Info, then the Keywords or IPTC section. The contains-synthetic-performer keyword is added in the Keywords field within the File Info panel. Photoshop’s export and Save As workflows generally preserve XMP metadata, but sellers using the Export As function or saving for web should verify that metadata preservation is enabled in the export settings — some web-optimized export profiles strip metadata by default.
Adobe Bridge and Lightroom
Adobe Bridge and Lightroom both support batch XMP metadata editing, making them appropriate for sellers managing large catalogs of AI-generated images. In Bridge, select the images, open the Metadata panel, and add the keyword in the IPTC Keywords field. Lightroom’s metadata fields work similarly. Batch application is the most practical approach for sellers who need to tag multiple images efficiently.
ExifTool (free, command line)
ExifTool is a free, open-source command-line tool that reads and writes XMP and EXIF metadata across all common image formats, available at exiftool.org. The command to add the contains-synthetic-performer keyword to a file’s dc:subject field is: exiftool -XMP-dc:Subject="contains-synthetic-performer" imagefile.jpg. ExifTool is the most reliable tool for batch processing large image libraries and for verifying that metadata is present after adding it. Sellers managing AI image workflows at scale should incorporate ExifTool into their image processing pipelines rather than manually tagging each file.
macOS Preview
macOS Preview supports XMP keyword tagging through the Inspector panel. Open the image in Preview, open the Inspector through the Tools menu, select the Keywords tab, click the plus button, and add contains-synthetic-performer exactly. Close and reopen the Inspector to verify the keyword is still present. Preview’s XMP support is adequate for individual files but is not practical for catalog-scale batch processing.
The Retroactivity Question: Existing Uploaded Images
Amazon’s stated position is that sellers are not required to add the metadata retroactively to content that was already published before the requirement was announced. This carve-out from Amazon’s own policy does not address the question of whether sellers have legal obligations under New York law for images uploaded between June 9, 2026 — when New York’s statute took effect — and July 27, 2026 — when Amazon’s requirement became effective.
The New York law gap period
New York General Business Law Section 396-b was in effect from June 9, 2026. Sellers who uploaded photorealistic AI-generated people in product images between June 9 and July 27 were subject to the New York disclosure obligation regardless of what Amazon’s policy said at that time. The New York statute does not exempt sellers from liability because their distribution platform had not yet implemented a disclosure mechanism. Sellers with significant volumes of AI lifestyle imagery uploaded during this six-week window should evaluate their exposure under the New York statute with legal counsel regardless of Amazon’s retroactivity carve-out.
Images uploaded before June 9, 2026
AI-generated lifestyle images uploaded before June 9, 2026 — before New York’s law was in effect — were not subject to the statutory disclosure obligation at the time of upload. Amazon’s policy does not require retroactive tagging of this content either. Sellers who uploaded AI imagery before June 9 and have not added the metadata tag are not currently in violation of either Amazon’s policy or the New York statute for that pre-June 9 content, under the current state of both frameworks.
Building Compliance Into Your Image Workflow
For sellers who regularly use AI-generated lifestyle imagery, treating the metadata requirement as a one-time catalog audit misses the ongoing obligation. Every new AI-generated image uploaded after July 27 requires the tag before upload. Building the requirement into the image creation and processing workflow is the only reliable way to maintain continuous compliance.
Where to insert the tagging step
The tagging step belongs immediately before the final file is prepared for upload — after all image processing, resizing, and format conversion steps are complete, but before the file is delivered to Amazon’s image upload system. Adding the tag at an earlier stage in the pipeline, then running the file through processing steps that strip metadata, produces a non-compliant final file even though the tag was applied at some point in the process. For teams using an image production workflow — designer, reviewer, uploader — the compliance verification step (confirming the tag is present in the final file) should sit with the uploader role, not the designer, to catch metadata stripping that occurred during file handoff.
Documenting AI-generated content
Maintaining a record of which images in a catalog contain AI-generated people — with the generation tool, generation date, and upload date recorded — provides documentation relevant to both Amazon compliance and the underlying legal exposure. When New York or California enforcement activity increases, or when the legal framework expands to other states, sellers with documentation of their compliance efforts are in a better position than those who cannot reconstruct what was AI-generated and when it was uploaded.
What Amazon Does When It Finds Non-Compliant Images
Amazon’s product image guide authorizes removal of images that violate image policies and suppression of listings from search when the listing lacks a compliant main image. Amazon has not published a specific enforcement timeline or penalty schedule for the AI metadata requirement distinct from its general image policy enforcement, but the general enforcement framework applies.
Image removal and listing suppression
Amazon’s standard image policy enforcement removes non-compliant images from listings. If the removed image is the listing’s main image — the first image in the gallery that appears in search results — the listing is suppressed from search until a compliant main image is provided. A suppressed main image stops organic search traffic immediately — ranking recovery after reinstatement can take days or weeks depending on the category and listing history. For sellers whose main image is an AI-generated lifestyle shot featuring a person without the required metadata tag, the suppression risk is the most immediate operational consequence of non-compliance.
Account health implications
Amazon’s listing quality enforcement framework treats image policy violations as listing quality events that can contribute to Account Health Rating degradation when they accumulate across multiple ASINs or recur after correction. A single non-compliant AI image on a single ASIN is a listing-level compliance matter — but a pattern across a catalog, particularly if Amazon’s systems detect it after notification was provided to the seller, may be treated as a more serious listing quality issue. See our Amazon Account Health Rating guide for the enforcement escalation framework.
The Legal Exposure Under New York and California Law
The legal exposure for Amazon sellers using AI-generated human likenesses without disclosure is not hypothetical. New York’s statute is in effect, carries specific civil penalties, and defines the disclosure obligation in terms that apply directly to seller-created product imagery.
New York General Business Law Section 396-b: who is liable
The statute places the disclosure obligation on the party that “produces or creates” the commercial advertisement. For Amazon product imagery, that is the seller — not Amazon as the platform and not the AI generation tool as the software provider. A seller who uses Midjourney, Adobe Firefly, or any other generative AI tool to create a photorealistic human lifestyle image, then uploads it to an Amazon product listing, has produced a commercial advertisement containing a synthetic performer and is subject to the disclosure obligation under Section 396-b. The exemption in the statute for platforms that “merely publish or distribute” the content applies to Amazon, not to sellers who created the content and chose to upload it.
Civil penalty structure
The New York statute’s civil penalty structure is $1,000 for a first violation and $5,000 for each subsequent violation. “Violation” likely means each separate advertisement — each product listing image — that fails the disclosure requirement. A seller with 50 AI-generated lifestyle images across their catalog, uploaded after June 9 without disclosure, potentially faces 50 separate violations — a first violation at $1,000 and 49 subsequent violations at $5,000 each, producing theoretical maximum exposure of $246,000 from the civil penalty structure alone, before any consideration of reputational harm or litigation costs. Actual enforcement actions to date have not tested the per-image violation interpretation, but the statutory language supports it.
Practical enforcement risk today
New York’s statute was enacted on June 9, 2026. No published enforcement actions under the statute have been reported as of August 7. The enforcement risk is currently theoretical rather than documented, but the statute is in effect, the obligation is real, and the window for proactive compliance is now — before enforcement actions create precedent that clarifies and potentially extends the statute’s reach. Sellers who address their AI image compliance now face a far more manageable situation than those who address it after a first enforcement action.
When Non-Compliance Creates Listing and Account Enforcement Risk
For most sellers, the practical risk from the AI image metadata requirement is Amazon listing enforcement rather than state law civil penalties. Understanding the specific pathways from image non-compliance to listing and account enforcement allows sellers to prioritize remediation correctly.
The main image suppression scenario
The highest-impact enforcement scenario is Amazon removing a non-compliant main image from a top-revenue ASIN, suppressing the listing from search until a compliant image is provided. For sellers whose main image is an AI-generated lifestyle shot, this is the immediate remediation priority — replace or tag the main image first, then address secondary gallery images and A+ Content. A suppressed main image loses organic ranking in real time. Recovery after reinstatement is not immediate and may require advertising support to accelerate ranking recovery.
The IP complaint risk from AI-generated human likenesses
A secondary enforcement risk arises when an AI-generated person bears a resemblance to a real identifiable individual — a Vine Voice, a public figure, or a person who can establish that the AI-generated image constitutes an unauthorized use of their likeness. Right of publicity claims under state law, including California and New York, protect individuals against unauthorized commercial use of their likeness — and AI-generated images that resemble real people may trigger right of publicity claims from those individuals. Amazon’s IP complaint system provides a mechanism for rights holders to report such claims, and an IP complaint from a person claiming their likeness was used without authorization in an AI-generated image can produce the same listing deactivation consequences as any other IP complaint. See our Amazon IP complaint guide for the IP complaint response framework.
Frequently Asked Questions About Amazon’s AI Image Metadata Requirement
Does this apply to AI-generated product images that don’t show people?
No. Amazon’s requirement specifically covers images containing photorealistic AI-generated people. A fully AI-generated product image showing only the product — no human figure, no hands, no body parts — is not covered by this specific requirement. Amazon’s general image accuracy requirements still apply to AI-generated product images regardless of whether they contain people: the image must accurately represent the product that ships. An AI-generated product image that makes the product look different from what actually ships creates an “item not as described” complaint risk independent of the AI disclosure requirement.
What if my AI-generated image shows only hands or partial body parts?
Amazon’s guidance focuses on “photorealistic AI-generated people” without defining the minimum human content that triggers the requirement. Hands alone, feet alone, or other isolated body parts shown in a product context — such as a hand holding the product — are in an ambiguous zone. The safest interpretation is to apply the tag when any AI-generated human body part is present in a photorealistic image, given that the purpose of the requirement is disclosure of AI-generated human content in commercial imagery. Applying the tag to borderline cases has no downside and eliminates compliance uncertainty.
My AI tool adds a watermark or notation to its output. Does that satisfy the requirement?
No. Amazon’s requirement specifically requires adding the contains-synthetic-performer keyword to the XMP dc:subject field in the image file’s metadata. A visible watermark on the image, a notation in the filename, or any disclosure mechanism other than the specific XMP metadata field does not satisfy Amazon’s technical requirement. Amazon reads the metadata field — not the visible content of the image — to identify compliant files. The tool’s own watermarking or notation is irrelevant to Amazon’s compliance determination.
We use AI to generate backgrounds and then composite real people into them. Does the requirement apply?
No. Amazon’s requirement applies to people who were entirely generated by AI — not to real people photographed in AI-generated environments. A composite image showing a real, photographed person in an AI-generated background does not trigger the contains-synthetic-performer requirement because the person is real, not AI-generated. The background’s AI origin is not what the requirement targets. If the same image contains both a real photographed person and a separate AI-generated human figure in the same scene, the presence of the AI-generated human figure would trigger the requirement regardless of the real person’s presence.
Amazon said we don’t need to retroactively tag existing content. Are we still at risk?
Amazon’s retroactivity exemption applies to Amazon’s own image policy enforcement — Amazon is not currently requiring sellers to add metadata to already-published content. The New York statute operates independently of Amazon’s policy. Images containing AI-generated people that were uploaded between June 9, 2026 (when New York’s law took effect) and July 27, 2026 (when Amazon’s requirement became effective) may have been non-compliant with the New York statute at the time of upload. Amazon’s retroactivity exemption does not affect that potential statutory liability. Sellers who uploaded significant volumes of AI lifestyle imagery during that six-week period should seek legal counsel to evaluate their specific exposure under the New York statute.
How DAM Law Firm Can Help
DAM Law Firm handles the legal and enforcement dimensions of Amazon listing compliance — including the situations where technical compliance failures produce Amazon enforcement consequences and where the underlying state law obligations create legal exposure beyond Amazon’s platform.
Listing enforcement from image compliance failures
When non-compliant AI images produce listing suppression, account health events, or account-level enforcement through Amazon’s image policy enforcement framework, our Amazon listing suspensions team and our Amazon account suspensions team handle reinstatement through the listing quality appeal process and, when needed, through pre-arbitration demand letters when Amazon’s enforcement response is disproportionate to the violation. Our Amazon product compliance lawyer team handles the listing content compliance review that identifies and addresses image policy issues before they produce enforcement events.
IP complaints from AI-generated human likenesses
When an AI-generated image produces a right of publicity claim or an IP complaint from a person claiming unauthorized use of their likeness, our Amazon IP complaints team handles the complaint response, the retraction request, and the listing reinstatement process alongside any direct legal response to the underlying right of publicity claim.
Legal exposure under New York and California law
For sellers who need to evaluate their exposure under New York General Business Law Section 396-b or California’s AI disclosure framework for AI imagery uploaded during or after the relevant statutory periods, our business law for sellers team handles the legal analysis of disclosure obligation compliance across the seller’s Amazon catalog and advises on remediation strategy that addresses both the Amazon policy requirement and the state law obligations simultaneously.
If AI image enforcement is affecting your Amazon listings, or if you need legal guidance on your obligations under the New York or California AI disclosure frameworks, contact our team for a same-day assessment.
Related DAM Law Firm services:
- Amazon Product Compliance Lawyer — listing content compliance review including image policy and AI disclosure requirements
- Amazon Listing Suspensions — listing reinstatement when image policy violations produce listing suppression or deactivation
- Amazon Intellectual Property Complaints — IP and right of publicity complaint response when AI-generated human likenesses generate infringement claims
- Amazon Account Suspensions — account reinstatement when image compliance issues escalate to account-level enforcement
- Business Law for Sellers — legal analysis of AI disclosure obligations under New York and California law for sellers using AI-generated imagery
This article is for general informational purposes only and does not constitute legal advice. Every situation depends on its specific facts, applicable law, and current Amazon policies. Contact DAM Law Firm for advice tailored to your situation.
Related articles from DAM Law Firm
- Amazon Product Title 75-Character Limit: Complete Seller Guide
- Amazon IP Complaint: When It’s False, Abusive, or Filed by a Competitor
- Amazon Account Health Rating: What It Measures and How to Protect It
- Amazon Inauthentic Item Suspension: Why Genuine Sellers Get Flagged and How to Appeal
- Amazon Seller Account Suspended: What to Do in the First 24 Hours
