Pinterest will fold its rules for generative AI tools into the contract every user accepts on November 12, 2026, when a revised Terms of Service adds the company's GenAI Acceptable Use Guidelines to the list of policies that carry contractual force. On the same date, an updated version of those guidelines adds a sentence that names the penalties for abuse - loss of access to Pinterest's generative AI features, suspension or termination of the account - and an updated Privacy Policy adds Pinterest's generative AI models to the models it trains on users' Pins, whenever those Pins were posted.

In Short

Pinterest's rules for its AI image and chat tools stop being a stand-alone help page on November 12, 2026, and become part of the legal agreement that every user accepts. Breaking them - for example by trying to jailbreak a tool, scrape it or use it to make sexual or misleading content - can now cost a user access to the AI features, or the account itself. At the same time, Pinterest says human staff may review what people type into its AI tools and what comes out, and its new privacy policy says user Pins train its generative AI models as well as its other machine learning systems.

One sentence that changes the status of the rules

The change in the Terms of Service is a single line. The opening section lists the policies that the Terms "include, and incorporate by reference" - documents whose breach counts as a breach of the user agreement itself. The version in force since April 30, 2025 names four: the Community Guidelines, the Advertising Guidelines and Merchant Guidelines, the Commercial and Branded Content Guidelines, and the Enforcement practices. The version effective November 12, 2026 adds a fifth: "Our GenAI Acceptable Use Guidelines, which explain how you may and may not use Pinterest's generative AI tools."

That matters because of what the Terms already allow. Section 7, unchanged, says: "We may terminate or suspend your right to access or use Pinterest for any violation of these Terms." Until November 12, the GenAI guidelines sit outside that sentence. The current guidelines state that use of the tools is subject to the Terms, the Privacy Policy and the Community Guidelines, so serious misuse could already be caught through those documents. From November 12, a breach of the AI rules is a breach of the Terms in its own right.

Acceptance is passive. Section 13(a) of both versions says Pinterest will notify users of material changes before an update takes effect and that "Your continued access to or use of Pinterest after such an update constitutes your binding acceptance of such changes." The Terms page carries a banner announcing the November 12 date. Pinterest gives no announcement date, but according to ConductAtlas, a service that archives platform policies, the banner was missing from its April 19, 2026 capture of the page and present in its September 2, 2026 capture.

What the guidelines allow and forbid

The GenAI Acceptable Use Guidelines are short. According to Pinterest's policy site, they cover any Pinterest generative AI product, feature or capability - which the page calls GenAI tools - and define use broadly, to include both what a user puts into a tool and what comes out of it. The page does not name a single product, and the current version shows no effective or last-updated date.

There is one permitted category. The guidelines encourage use of the tools for inspiration that leads to action, broken into three examples: exploring ideas, finding a personal style and bringing creative visions to life.

The prohibited list is far longer. It runs to 17 items in three groups, according to the guidelines page.

Dangerous or illegal use

Eight bans sit in the first group. Users may not use the tools to violate intellectual property or privacy rights or other laws. They may not compromise systems or networks or get around safety and security measures, with jailbreaking given as the example. Reverse engineering and data scraping are banned, as is collecting, sharing or storing personal, sensitive or confidential information without the right to do so. The group also prohibits encouraging self-harm or harm to others, activity that risks imminent and serious physical or emotional harm, human or wildlife trafficking, and promoting or manufacturing dangerous substances and weapons.

Misleading or deceptive use

Three bans make up the second group: impersonating any person or organisation or misrepresenting an affiliation with one; creating or spreading harmful misinformation, disinformation or prohibited campaigns; and spam, scams or other deceptive practices, with malware and phishing as examples.

Explicit, hateful or violent use

The third group has six bans. The tools may not be used to create sexual content or visible intimate body parts, to perpetuate intimate image abuse such as non-consensual intimate images or sextortion, or to compromise the safety of minors. Hateful, attacking, shaming or threatening activity is banned, along with encouraging acts or threats of violence and supporting violent actors.

Each group ends with a cross-reference to the Community Guidelines, and the second group also points to the Advertising Guidelines. The guidelines add that the generative tools are subject to the same safety restrictions as the rest of Pinterest's products, features and content.

The penalties, now in writing

The November 12 version of the guidelines is almost identical to the current one. Pinterest's preview of the updated page shows two differences, and only one carries weight. The paragraph that ties the guidelines to the Terms, the Privacy Policy and the Community Guidelines gains a sentence stating that abuse of Pinterest's generative AI technology may result in account enforcement. It names three outcomes: loss of access to the generative AI features, account termination and account suspension.

The second change is a relabelled cross-reference. In the footnote to the misleading-use group, a link previously described as covering prohibited and regulated products or practices now refers to dangerous goods and activities. The link destinations are unchanged.

Every one of the 17 bans, the permitted uses and the human review paragraph read the same in both versions. In other words, Pinterest has not tightened the AI rules; it has made them part of the contract and spelled out what breaking them costs.

Human review of prompts and outputs

One paragraph in the guidelines, unchanged in the update, deals with quality and safety. It says Pinterest's generative tools aim to be helpful but can be imperfect, and that to improve its services Pinterest's human agents may review user content - including inputs and outputs - along with other information related to use of the tools. The page does not mention any way for users to opt out of that review.

The incorporation gives that paragraph a new setting. A user who keeps using Pinterest after November 12 accepts Terms that incorporate guidelines which describe human review of prompts. Pinterest's Help Center page on AI at Pinterest says separately that the company uses human reviewers and public feedback, where appropriate, to refine its AI.

The scope is layered. The guidelines say they apply on top of the Terms, the Privacy Policy and the Community Guidelines, and, where relevant, Pinterest's Business AI terms, Merchant Guidelines, Advertising Guidelines, Commercial and Branded Content Guidelines and Developer Guidelines. Advertisers using Pinterest's generative ad features therefore face several rulebooks at once. The revised Advertising Guidelines, which add bans on synthetic-media misrepresentation, scam tooling and impersonation, take effect on the same November 12 date, after a notice Pinterest emailed to advertisers on September 4, 2026.

The privacy side: generative AI models join the training language

The Privacy Policy changes on November 12 as well, and its generative AI passages shift in three places, according to Pinterest's published preview of the new policy.

First, the section on information Pinterest collects adds interactions with features supported by generative artificial intelligence technology. Second, the sentence on how Pinterest uses information to train its technology now covers generative AI explicitly. The current policy says Pinterest uses information to train, develop and improve its technology, such as its machine learning models, regardless of when Pins were posted. The new version keeps the reference to Pins of any age and extends that training to Pinterest's generative artificial intelligence models. Third, in the section that sets out legal bases for users in the EEA, the part on showing relevant content gains a point about using machine learning and generative artificial intelligence to do so.

Other generative AI passages do not change. Both versions say Pinterest offers generative AI features that let advertisers add limited content to their ads. The summary of changes at the top of the new policy speaks of covering additional Pinterest services and giving more detail on how Pinterest may use data, including to keep the platform safe. The summary on the current policy, by contrast, said Pinterest had added more detail on how it trains the models behind its products, including generative AI ones; that language described the previous round of changes.

The Terms of Service connect the two documents for users outside Europe. A new sentence, also effective November 12, reads: "If you reside outside of the EEA, Switzerland and the United Kingdom, your acceptance of these Terms constitutes your consent to the processing activities described in our Privacy Policy under the laws of your jurisdiction." For those users, accepting the Terms becomes consent to processing that includes training generative AI models on Pins.

Pinterest's defaults were already a subject of scrutiny. A May 2026 Surfshark study found that 8 of 10 apps enrolled users in AI training by default, with Pinterest among them and its opt-out taking three to five actions in the app, as PPC Land reported when Meta dropped an Instagram AI tagging tool in July 2026.

Which tools the rules cover

The guidelines name no products, but Pinterest's Help Center page on AI at Pinterest lists several. Pinterest Canvas is described as an image-generation foundation model that creates backgrounds for product images in ads; according to the page, data was first used in its development in or around January 2023, and it was trained on more than 600 million data points. Navigator is described as a family of multimodal foundation models that produce text outputs, including for conversational features; the page says training with synthetic data began in January 2026. The page also lists generative AI support tools that give users help and guidance.

PPC Land has followed those tools as they reached users and advertisers. Pinterest introduced Performance+ creative in October 2024, using generative AI to turn flat product backgrounds into lifestyle scenes; in a test by Walgreens on about 50 products, Pinterest reported a 55% higher click-through rate and a 13% lower cost per click. In June 2025 it launched AI auto-collages built from product catalogues. By the first quarter of 2026, Canvas had gained real-time editing and Pinterest Assistant, which entered beta in late 2025, was in a controlled rollout, according to PPC Land's analysis of the Q1 2026 results. Pinterest Assistant is limited to users aged 18 and over, a point the company made when it ruled out companion AI as California enacted Adam's Law in September 2026.

Labelling is the other half of the system. Pinterest began showing an "AI modified" label on April 30, 2025, using IPTC metadata together with its own classifiers. The Help Center page describes Gen AI labels on image Pins when the content owner or Pinterest's detection systems indicate that an image was generated or altered by AI. In October 2025 Pinterest added controls that let users see less AI-generated content in beauty, art, fashion and home decor.

What the Terms still do not say

For all the attention to generative AI, the Terms of Service themselves remain silent on several points that other platforms have recently addressed. There is no clause granting Pinterest a licence to train AI on user content - the licence in Section 3(b) over user-generated content is word for word the same, though it already covers use, modification, derivative works and monetisation by Pinterest, its affiliates and service providers. Model training is handled in the Privacy Policy, not the contract. There are no rules on AI agents acting for users, and no ban on prompt injection.

The scraping ban in the Terms is also unchanged. Section 2(a) still bars users from scraping, collecting or copying data from Pinterest "by using automated means (without our express prior permission)". The GenAI guidelines' own scraping ban adds a second layer, aimed at the AI tools rather than the site.

The same November 12 revision makes other changes outside the AI field. It moves user arbitration outside Europe from the American Arbitration Association to JAMS, gathers 25 or more similar claims filed within 90 days into a single arbitration, lets arbitrators split fees with claimants' lawyers and funders, and adds a 30-day opt-out from arbitration. It also adds a sentence binding users to YouTube's Terms of Service when they access YouTube content on Pinterest.

Why the AI rules matter for marketers

For advertisers, several of Pinterest's generative tools sit inside the ad-buying workflow, from Performance+ creative and Canvas backgrounds to auto-collages. The guidelines' bans on impersonation, misleading content and intellectual property violations apply to outputs as well as inputs, and from November 12 they sit alongside revised Advertising Guidelines that address synthetic media directly.

Disclosure is becoming a commercial question in its own right. The IAB's voluntary AI disclosure framework, issued in August 2026, cited an NYU Stern finding that disclosing generative AI involvement cut ad click-through by 31.5%. In October 2026 Resemble AI reported that only 2 of 3,379 AI-flagged Meta ad files disclosed AI use. The GenAI guidelines say nothing about labelling AI-made creative; on the platform, labels rely on metadata and Pinterest's own detection, according to the Help Center.

The privacy change matters for creators and brands that publish organic Pins. The new policy text says Pins, regardless of when they were posted, train Pinterest's generative AI models, and the Terms treat acceptance as consent outside Europe. A brand's back catalogue of Pins therefore falls within the training language as written. Pinterest's commitment to AI infrastructure is substantial: in June 2026 it committed $4 billion to AWS through 2031, with Trainium chips for model training, after cutting about 780 jobs in January 2026 as investors questioned its AI pivot.

The approach differs from Meta's. When Meta revised its commercial terms in October 2026, it wrote new rights over business content into the contract, including crawling advertisers' websites for AI training. Pinterest's contract changes go the other way: they bind users to rules on how they may use Pinterest's AI, while the rights Pinterest takes over user data for AI are described in the Privacy Policy. Will the penalties be used? The guidelines list what can happen, but Pinterest has not published enforcement figures for its generative tools.

Timeline

Summary

Who: Pinterest and its users, including advertisers and creators who use its generative AI tools such as Canvas-powered ad backgrounds, auto-collages and Pinterest Assistant. The consent provision applies to users outside the EEA, Switzerland and the UK.

What: Pinterest's revised Terms of Service incorporate the GenAI Acceptable Use Guidelines, making their 17 prohibited uses, from jailbreaking and scraping to sexual and deceptive content, part of the user contract. The updated guidelines add that abuse can lead to loss of AI features, suspension or termination. The guidelines keep a paragraph allowing human review of inputs and outputs, and the revised Privacy Policy extends training on Pins, whenever posted, to Pinterest's generative AI models.

When: The changes take effect on November 12, 2026. The Terms being replaced have been in force since April 30, 2025, and the update notice was on the Terms page by September 2, 2026, according to ConductAtlas.

Where: The AI rules apply to Pinterest users worldwide. The provision treating acceptance of the Terms as consent to Privacy Policy processing applies outside the EEA, Switzerland and the UK.

Why: Pinterest does not explain the change in the Terms. The effect is to give its AI usage rules contractual force and named penalties as generative tools spread through its consumer and advertising products, while the Privacy Policy update describes generative AI training on user Pins more explicitly.