Pinterest see-fewer AI Pins: beauty, decor, fashion, food and drink
Pinterest's see-fewer GenAI controls cover beauty, art, fashion, decor, health and sport, plus food and drink from December 12, 2025. What it means for Pins.

Pinterest's see-fewer GenAI controls, launched October 16, 2025, let people turn down AI content in categories including beauty, art, fashion, home decor, health and sport, and food and drink was added on December 12, 2025. If you publish generated images in those categories, expect some viewers to have asked not to see them.
Dates and categories
The Pinterest newsroom post gives the facts below, read on 2026-10-03.
| Item | What the page says |
|---|---|
| Launch | October 16, 2025 (Android, desktop; iOS rolling out) |
| Categories at launch | Beauty, art, fashion, home decor, health, sport |
| Added later | Food and drink, on December 12, 2025 |
What the page does not say
The post describes the controls. It does not give a figure for how many people use them, how much reach a Pin loses, or how Pinterest decides that a Pin is AI content for this purpose. This post does not guess at any of that.
Pinterest's separate help page on gen AI labels says it labels "AI modified" content using IPTC metadata, classifiers and creator disclosure. The pages read do not tie the labels to the see-fewer controls, so treat them as related and not identical.
Category risk for generated images
A rough way to sort a content calendar, using only the category list above:
| Category on the list | Typical generated Pin | Plan |
|---|---|---|
| Beauty | Face or skin close-ups | Prefer real photos of the real product |
| Home decor | Rooms and styling | Mix generated scenes with real room photos |
| Fashion | Outfits on people | Show the real garment, not an invented one |
| Food and drink | Plated dishes | Show the dish that the recipe actually makes |
| Art, health, sport | Illustration, wellness, action | Label and be specific about what is real |
Make generated Pins that earn the click
If a viewer has asked for less AI content, a Pin that depends on looking photographic is the weakest fit. Pins whose value is the idea, a recipe, a layout or a how-to, hold up better because the image is not the claim. Add specific text to the Pin so the content stands on its own.
The Image API supports input_references, so a real photo can anchor a generated variation, and aspect_ratio can be set per request, including 2:3 for tall images. Check the model's descriptors for the ratios it accepts.
{
"model": "openai/gpt-image-2.5",
"prompt": "A tall styled shelf scene around this real vase, warm daylight, no text",
"input_references": [
{ "type": "image_url", "image_url": { "url": "https://example.com/real-vase.jpg" } }
],
"aspect_ratio": "2:3"
}Disclose when it applies
Where Pinterest's label rules apply to your Pin, use the disclosure option instead of hoping detection misses it. A clear disclosure is cheaper than an appeal.
Sources
Related posts
More in Use cases
- Planned maintenance notice video: preview the frame, then render
Send a 20-second planned-maintenance video with a Sume avatar: create a preview, check the frame, then call generate-video once. Cost by tier included.
- Planter and plant pot photos with AI: scale, drainage and plant
Staging a planter with a plant in it changes how big the pot looks. Sume image API prompt and checks for scale, drainage hole, saucer and the plant itself.
- Playable AI worlds vs pre-rendered branches on Sume
PixVerse R2 is a real-time world model with WASD. Sume has no live session; pre-render branches with start and end frames and splice them in Timeline.
- Podcast episode art: 12 numbered cards from one prompt skeleton
Make a season of consistent episode art with number and title drawn in. Ideogram 4.5 lists $0.06 per card at medium, so twelve cards are $0.72 at list.
Written by Sume