
Pinterest's catalog documentation represents AI disclosure as an association between an asset URL and its disclosure values. A product containing a real photograph, an AI-modified scene and a synthetic-person video therefore needs an accurate asset-level record. A generic note saying the product uses AI does not establish that each submitted URL has been mapped correctly.
Key Takeaways
- Identify which exact image or video has which AI involvement.
- Use the supported catalog representation, not a guessed global switch.
- Embedded provenance, feed disclosure and product accuracy remain separate checks.
Inventory the actual mixed-media set
The official retail-catalog specification describes ai_disclosures for realistic imagery, video or generated humans that could be confused with a real person or event. Its example associates a URL with a disclosure array containing values such as ai_modified and synthetic_performer.
Start from the files actually being distributed. Record which contain meaningful AI modifications and which contain a synthetic performer. Do not assign a human-performer classification to an ordinary product-only photograph simply because another asset in the same product uses a generated model.
Use the SKU-to-image record to keep the commerce identity intact alongside the source classification. A correct disclosure on the wrong variant is still a faulty submission.
Preserve the URL-to-disclosure relationship
For a proposed apparel set, the studio might have an approved real garment photograph, an AI-edited room scene and a clip featuring a generated person. The disclosure ledger should describe each actual file, not infer all of their origins from the garment's product identifier.
| Ledger field | Purpose |
|---|---|
| Asset and revision | Identifies the exact media reviewed |
| Submitted URL | Connects the file to the feed record |
| AI involvement | Records the truthful production account |
| Supported disclosure values | Supplies the relevant catalog classification |
When replacing an asset, update the relationship deliberately. A new URL or a changed production process can make an old mapping inaccurate even when the product title stays the same.
Use the integration's supported serialization
Have the responsible catalog maintainer inspect the final outgoing field using the current feed or connector format. The official example shows the conceptual URL and disclosure association; it does not prove that every connector exposes the same user interface.
Do not silently omit the field because an integration lacks a visible control. Investigate its current support and document any unresolved limitation. Likewise, do not invent a parameter or change the contents of a signed media address to carry disclosure text.
The AI provenance guide explains metadata embedded within files. The existence of such metadata does not, by itself, prove that this separate Pinterest feed field was supplied.
Review the real offer and processed result
Disclosure does not make an inaccurate product depiction acceptable. Compare garment structure, color, included items and claims with the real source before distribution. Avoid implying that a synthetic performer is an actual customer testimonial or verified event.
The Product Pin guide explains the independent need for an accurate commerce record and retailer destination. After an authorized submission, inspect processing feedback and any observable resulting treatment. Record what was actually verified. Do not promise a specific visible badge, legal compliance or distribution outcome merely because an outgoing field is present.
Related tasks
- How Do I Keep AI Image Provenance Metadata Through Editing and Delivery?
- Does Adding a Price to My AI Product Poster Create a Pinterest Product Pin?
- How Do I Keep AI Product Images Matched to the Correct SKU?
Sources and scope
- Pinterest: Get started with retail catalogs: The ai_disclosures field pairs an asset URL with disclosure values; its example includes ai_modified and synthetic_performer for realistic AI images, videos or humans that could be confused with a real person or event.
Based on the complete official catalog source captured October 29, 2026. No feed, label, asset, API or displayed Pin was tested. The mixed-asset ledger is proposed. The documented lowercase example values are retained; no unsupported legal-compliance conclusion, universal visible-watermark rule or automatic metadata-to-feed mapping is claimed.
Prepared by Imagild Editorial, the company editorial team responsible for these guides. AI assisted research and drafting. This is documentation-based guidance with proposed examples, not a claim of hands-on testing or an independent product review. Report a correction.
Cover: AI-generated conceptual editorial illustration; not a real application screenshot, observed output or before-and-after test.