Does Pinterest train AI on your Pins? The setting and what it misses

The Morse teamPublished October 28, 202611 min read

Yes, as of October 28, 2026, Pinterest trains Pinterest Canvas, its image generation model, on the Pins people save or upload, and Pinterest's help page treats adults as included unless they opt out by clearing the box Use your data to train Pinterest Canvas under Settings, then Privacy and data. Pinterest opts out users under 18 automatically and doesn't let them opt in. That box covers one model. Pinterest's privacy policy also lets it train its other machine learning models on your information "regardless of when Pins were posted," and a help page rewritten this month says its Navigator models are trained on public content and chat logs, with no setting named for them.

We read Pinterest's privacy policy (effective April 30, 2025) and the version that takes effect November 12, 2026, its Terms of Service, its GenAI Acceptable Use Guidelines, nine Help Center articles on AI, privacy, and feed settings, and two newsroom posts, all on October 28, 2026. Every fact below links to its source. Pinterest deserves credit for one thing. It publishes a description of what each of its AI models was trained on, with counts and start dates, which most social apps don't.

Pinterest's AI training settings at a glance

Read on October 28, 2026.

Does Pinterest use your images to train AI?

Yes, for Pinterest Canvas and for its other models. Pinterest's AI at Pinterest page says Canvas "has been trained on more than 600,000,000 data points, composed of publicly available Pins, Pin descriptions, and Pin metadata that Pinterest users have saved or uploaded." The same page says the data was collected "as early as 2009," was first used for Canvas "in or around January 2023," and that "the data collection is ongoing."

Pinterest's own example of Canvas at work is for advertisers. Its GenAI settings page says Canvas makes "ads more engaging for advertisers by enabling the creation of backgrounds on product images." So a photo of your living room can help teach the model that draws a living room behind someone else's sofa ad.

Canvas isn't the only model. The privacy policy says Pinterest uses information "to train, develop and improve our technology such as our machine learning models, regardless of when Pins were posted." It names Pinterest's body type technology, "which is trained on images in Pins posted to our Services," as one example. For how other apps compare, see which AI companies train on your data.

How do you opt out of Pinterest AI training?

Clear one box on the web, or one toggle in the app. Pinterest's GenAI settings page gives these steps.

On the web:

1. Log in and click the More options icon at the bottom left of the screen.
2. Click Settings, then select Privacy and data.
3. Uncheck the box next to Use your data to train Pinterest Canvas.
4. Click Save.

In the Pinterest app:

1. Tap your profile picture at the bottom-right corner.
2. Tap your profile photo at the top-left of the screen.
3. Tap Privacy and data, then turn off the toggle next to Use your data to train Pinterest Canvas.

On a business account, Pinterest's privacy and data settings page says to tap your profile photo at the bottom right, then the ellipsis icon at the top right, to reach the same settings.

Site owners have a separate route. Pinterest's Canvas Opt Out page lets "content owners" opt their websites out "from future Pinterest Canvas foundation model training." You need a claimed website and three to five sample Pin IDs from your business, and Pinterest says an approved request comes with "additional Terms and Conditions."

Is Pinterest's AI training on by default?

For adults, yes, going by Pinterest's own wording. The GenAI settings page never states a default for adults. It says Pinterest "may use" what you save or upload to improve Canvas and that "you can opt out of this use of your data at any time." The one default it prints is for teens: "If you're under 18, you'll automatically be opted out of this setting and unable to opt in."

Pinterest's help pages give no different default for Europe. The British English version of the GenAI settings page has the same steps and the same under-18 rule, with no mention of the EU or the UK. Pinterest does print regional defaults for ads. Its privacy and data settings page says EU users are "automatically opted out" of Activity for ads reporting and Ads off Pinterest, and that minors in the EU, Switzerland, and the UK are opted out of "all ads personalization settings."

Europeans have one more route. The privacy policy says people in the EEA, Switzerland, and the UK can object "where features supported by generative artificial intelligence are based on legitimate interests," through a form to Pinterest's Data Protection Officer. Pinterest Europe Ltd., in Dublin, and Pinterest, Inc. are joint controllers for those users.

What Pinterest's AI models are trained on, and which have a setting

Pinterest's help pages and privacy policy describe four kinds of model. Only one has a switch you can reach. Read on October 28, 2026:

The Navigator row is new. When the Internet Archive saved the AI at Pinterest page on October 11, 2026, it described a single model, Navigator-1, trained on about 150,000 synthetic conversations, and said that data "was not collected from users" and "does not include personal information." The page we read on October 28, 2026 describes a family of Navigator models built from synthetic data plus public content and usage data. It says collection of content "dates to the launch of Pinterest in 2010." Pinterest's page dates the collection of usage data from AI features to "after October 2026," even though it already counts more than 100,000 conversational logs. Many guides treat the Canvas box as the whole opt-out and don't mention Navigator.

The same October 11 snapshot put Canvas at "approximately 500,000,000 rows." Today's page says more than 600 million data points.

Are secret boards used to train Pinterest's AI?

Pinterest doesn't say. Its Canvas description covers "publicly available Pins," and a secret board is one "that only you (and anyone you invite) can view." But Pinterest doesn't define "publicly available," and its privacy policy's training clause makes no split between public and secret Pins. For body type search, the policy says "Pinterest uses any Pins on Pinterest."

A secret board hides your collection from other people. A Pin you save to it from someone else's public board is still public on theirs. A private profile keeps you out of Pinterest search and search engines, by the same help page, and Pinterest doesn't say it changes training either. The only sure way to keep a photo out of Pinterest's training is not to upload it.

What changed in Pinterest's AI training policy?

The rules have widened in steps since 2023. From Pinterest's own pages:

• January 2023: Pinterest first used its Pin dataset to develop Canvas (Pinterest).
• April 30, 2025: the current privacy policy took effect. Its summary says Pinterest "added more details on how we train the models," and its training clause reaches Pins "regardless of when Pins were posted."
• April 30, 2025: Pinterest introduced Gen AI labels and an experiment to see fewer AI Pins, starting with categories like beauty and art.
• October 16, 2025: Pinterest rolled out controls to see less GenAI content, on Android and desktop first, in categories including architecture, art, beauty, and entertainment. A December 12, 2025 update added food and drink.
• January 2026: Navigator training on synthetic data began (Pinterest).
• Between October 11 and October 28, 2026: the AI at Pinterest page added public content and AI feature usage data to Navigator's training sources.
• November 12, 2026: the updated privacy policy takes effect. It names generative AI outright: Pinterest may use "information (such as Pins, regardless of when they were posted) to train, develop and improve our technology such as our machine learning models and our generative artificial intelligence models." It also says Pinterest collects information when you interact with "features that are supported by generative artificial intelligence technology."

How do you see fewer AI Pins on Pinterest?

Two controls change your feed. Pinterest describes both as feed settings, and neither is the training setting. The first is on a Pin. Pinterest's see fewer AI Pins page says to click or tap the ellipsis icon at the bottom right of a Pin in your feed and pick the "see fewer" AI Pins option, which shows only on Pins in certain categories.

The second covers the whole feed. On the web, click the more-options icon at the bottom left, select Refine your recommendations, and open the AI content tab. In the app, tap your profile picture, then the more-options icon at the top right, then Refine your recommendations (Pinterest). Turning off Show AI content turns off every AI topic, and Manage AI content by topic lets you pick. Pinterest says these settings "won't affect the ads you see on your home feed."

The AI modified label is how Pinterest marks what it finds. Its Gen AI labels page says the label shows at the bottom left when you open a Pin in close-up, and that Pinterest reads a Pin's metadata under the IPTC standard and runs classifiers that detect AI content "even if the content doesn't have obvious markers." A Pin's owner can also mark it. On ads, the label sits behind the ellipsis icon instead.

How Morse keeps your ideas out of AI training

Every model in the table above learns from a library Pinterest keeps: Pins uploaded since 2009, and the conversations people have with its AI features. A box in Settings takes your Pins out of one model, while the library stays on Pinterest's servers for the next one, under the license you granted when you posted.

Morse is built without that library. Morse doesn't train AI on your content, sell your data, or show ads. Your chat history is stored only in your browser, encrypted, not on Morse's servers. Studio, Morse's image generator, keeps the pictures you make in that same encrypted browser storage (privacy policy). There's no Canvas-style checkbox to hunt for before the next policy date, and Morse's servers keep no chat logs, so there's nothing like Navigator's 100,000 conversational logs to train on. If you clear your browser, that history is gone and Morse has no copy to restore, so keep the photos and boards you care about wherever you store them now.

Much of what ends up on a board starts as a question: what fits a narrow hallway, which paint goes with oak floors, how to plant a garden on a slope. Each Morse chat runs in a privacy mode you pick:

• Private, for the room plans and wish lists you'd otherwise pin to a secret board. Every new chat opens in this mode. Grok and open-weight models such as DeepSeek, Kimi, and GLM run on hosts under Morse's zero data retention agreements with its inference, search, and other providers, so they have agreed to keep nothing, your floor plan included. When a host confirms zero retention on a reply, Morse checks that confirmation and turns away any reply without it.
• E2EE, on Pro, for the board you'd never make public, like a nursery before you've told anyone. Before the message leaves your device, your browser verifies the enclave's hardware attestation and seals the text for that one enclave, so what passes through Morse's servers is ciphertext they can't open. The guarantee depends on the enclave hardware and attestation working as designed. Those chats run open-weight models and don't generate images.
• Anonymous, for asking GPT, Claude, or Gemini about a decorating style without your name or account attached. Morse's proxy strips your account, email, and IP address and sends the question under Morse's own account. The provider reads the question without knowing it's you, and may keep it for a limited time to check for abuse, under its own policy, so describe the room rather than give your address.

Signing out locks your history, and unlocking it takes two keys: one that never leaves your device, and one Morse's server gives you when you sign in. Because we don't store chats, we usually have very little to give authorities.

What to change on Pinterest today

• Your own Pins: uncheck Use your data to train Pinterest Canvas and click Save, on the web and in the app if you use both.
• Your website: file the Canvas Opt Out form if you've claimed your site. It's separate from the account setting.
• In the EEA, Switzerland, or the UK: use the privacy policy's objection form for generative AI features as well as the toggle.
• Your feed: turn off Show AI content under Refine your recommendations, and use "see fewer" on AI Pins you scroll past.
• Ideas you'd rather keep out of training: plan them in a Morse chat, since Morse doesn't train AI on what you write or make. For how other image apps handle this, compare Canva's settings, Adobe's terms, and Google Photos' promise.

Ask your first room-planning question from the Morse home page, no account needed, or create a free account with no card. Free includes 3 Studio images a day, and Pro adds E2EE chats for $19 a month.