Shutterstock states a Contributor Fund with first AI dataset earnings in Q4 2022, Getty Images states recurring compensation in its 10-K for fiscal 2025, and Wirestock states a 50 % commission per license (all checked 2026-11-09). None publishes a rate.
The record is thin for Adobe Stock and Pond5, so this page states no terms for them. It sets out each library's own wording and filings as one part of AI training data licensing.
How a Stock Library Licenses Its Collection for AI Training
A stock library licenses to AI developers the works it holds under contributor agreements, so the library signs the AI deal and the contributor is paid, if at all, under the library's own fund or royalty terms.
What is licensed is images, video and metadata such as captions. Shutterstock, Inc. offers content and metadata for AI model training through its Data, Distribution, and Services offering, and Getty Images Holdings, Inc. (Getty Images, iStock, Unsplash) offers data access and licensing.
The contributors' grant to the library for AI use is not in the record for any large library.
Some developers say they train only on licensed library data: Bria states that its models are trained exclusively on licensed data from Getty Images, Alamy and Envato (a marketing claim, checked 2026-11-09). Per-program image terms are covered in selling photos and images for AI training.
Training Datasets and Display Licenses Signed by Libraries
Library deals split into training datasets and display licenses, and the parties' own sources say which: Shutterstock's deals supply training datasets, while Getty's two dated AI deals in the record grant display. The table lists stock-library deals as of 2026-11-17, and no value is disclosed in any row.
| Date | Library | AI developer | Use | Stated scope (party's words) | Source |
|---|---|---|---|---|---|
| 2022-10-25 | Shutterstock | OpenAI (and LG) | Training datasets | expanded partnership announced with the Contributor Fund; "data deals" earnings from OpenAI datasets first issued Q4 2022; start of the original license not dated | Shutterstock contributor FAQ |
| 2025-10-31 | Getty Images | Perplexity | Display | global multi-year license for display across Perplexity's AI search and discovery tools | Getty newsroom |
| 2026-03-19 | Shutterstock | OpenAI, Black Forest Labs, Runway, ElevenLabs | Training datasets | named among companies powered by its licensed training datasets | Shutterstock release |
| 2026-06-21 | Getty Images | OpenAI | Display | multi-year display in ChatGPT search and discovery | Getty 8-K Ex. 99.1 |
"Display" means the parties state display; it does not mean training is forbidden. Neither Getty deal states training either way; see Getty Images AI licensing deals.
Contributor Funds, Bonuses and Commissions Compared by Library
Libraries pay contributors for AI use in three ways: a pooled fund shared when work enters a licensed dataset, a commission on each license, or a fixed payment per accepted asset. The table gives each library's own statement as of 2026-11-17, alphabetical, with the check date of each source.
| Library or platform | Mechanism | What the library states | Rate or formula | Opt-out or choice per dataset | Source |
|---|---|---|---|---|---|
| Adobe Stock | contributor bonus (Firefly) | terms page did not load (HTTP 403) on 2026-11-09, so no terms are read | not stated here | not stated here | helpx.adobe.com |
| Getty Images (incl. iStock) | compensation for training use; recurring basis | 10-K FY2025, filed 2026-03-16 | not published | not in the record | Getty 10-K FY2025 |
| Pond5 | not in the record | not stated | not stated | not stated | n/a |
| Shutterstock | Contributor Fund announced 2022-10-25; automatic for included contributors; first AI dataset earnings Q4 2022 | contributor FAQ, checked 2026-11-09 | not published | inclusion described as automatic; no choice per dataset described | submit.shutterstock.com |
| Wirestock | 50 % commission per license; fixed payment per approved asset, paid "instantly" | blog post dated 2026-07-13, checked 2026-11-09 | 50 % (commission projects) | license terms not described | wirestock.io |
Pooled Funds Paid When Work Enters a Dataset
Shutterstock and Getty Images both state that they compensate contributors whose work was used for AI training, and neither publishes a rate.
Shutterstock announced its Contributor Fund on 2022-10-25 and states that it "will directly compensate contributors if their IP was used" in developing generative models; inclusion is automatic and the first AI dataset earnings were issued in Q4 2022 (checked 2026-11-09). Fund detail sits on the Shutterstock data licensing profile.
The Fund page leaves three things open:
- the formula that divides money among contributors;
- the payment schedule;
- which datasets include which work.
Getty Images states in its 10-K filed 2026-03-16 that its generative AI image tool was trained on Getty content "while compensating contributors for the use of their copyrighted works as training data", and that models built with technology companies "compensate creators on a recurring basis".
It also states royalties of over USD 220 million paid to contributors in 2025 across all products, not an AI figure. More is on the Getty Images data licensing profile.
Per-License Commissions and Fixed Payments
Wirestock pays per license or per accepted asset instead of from a pool, and it offers creators' work to AI labs rather than licensing a library of its own. Its payment post, dated 2026-07-13 and checked 2026-11-09, states two project types:
- commission-based projects: "You earn 50% commission each time someone licenses one of your assets";
- fixed-payment projects: a payment per approved asset, paid "instantly".
The post does not describe license terms. Wirestock handles the work as an intermediary, and its terms are set out on the Wirestock profile and program terms.
Library Terms That Are Not on the Record
Two large contributor libraries, Adobe Stock and Pond5, have no AI compensation terms in the record this page relies on, so no rate, bonus formula or date is stated for them.
For Pond5, no AI licensing or contributor term is in the record. Adobe Stock names a contributor bonus tied to Firefly, but its contributor FAQ returned HTTP 403 on 2026-11-09 and 2026-11-16, so no amount or condition is stated. What Adobe says about its training data is covered under Adobe and Firefly training data.
What a Contributor Cannot See in a Library's AI Terms
A contributor can usually see that a library licenses for AI and that some payment exists, but not how much of a deal reaches contributors or which datasets contain their work. Four things are missing from the record:
- the formula and the share of deal income that goes to contributors;
- which datasets hold which works;
- the buyer of each dataset;
- the license term and whether delivered copies are deleted.
Shutterstock's page states no formula or license terms, Getty states no rate, and filings give totals only.
Can a Contributor Keep Work Out of a Library's AI Datasets?
No library page in the record describes a per-contributor opt-out from AI datasets, and Shutterstock describes inclusion in its Fund as automatic. The same holds for Getty Images, iStock, Adobe Stock and Pond5.
Platform settings elsewhere are described under opting out of AI training and its limits. Three routes remain open to a contributor, each limited by the agreement:
- ask the library in writing whether the work is in an AI dataset and whether the contributor can be excluded;
- withdraw works from the library where the contributor agreement allows it;
- reserve rights on copies hosted on the contributor's own site.
Whether a copy already delivered to an AI developer is deleted is a contract question, and no library page in the record answers it.
Directive (EU) 2019/790, in force since 2019-06-06, requires a rightholder's reservation against text and data mining of online content to be machine-readable (Article 4(3)).
Section 29A of the UK Copyright, Designs and Patents Act 1988 covers non-commercial research only, so commercial training there needs a license. Takedown routes are covered under removing content from AI training datasets.
What Library Filings Show About AI Licensing Income
Shutterstock and Getty Images are the only stock libraries in the record whose filings put numbers on data licensing, and neither files an AI-only revenue line. The table gives each filer's own figures as of 2026-11-17, in USD thousands unless stated.
| Filer and line | Period | Value | Note | Source |
|---|---|---|---|---|
| Shutterstock, Data, Distribution, and Services revenue | FY2023 | 137,323 | includes metadata licensing for AI plus Giphy and Studios; no AI-only line | 10-K FY2025 |
| Shutterstock, DD&S revenue | FY2024 | 175,251 | same scope | 10-K FY2025 |
| Shutterstock, DD&S revenue | FY2025 | 203,264 | up 16 % on 2024 | 10-K FY2025 |
| Shutterstock, DD&S revenue | Q3 2026 | 15,823 | Q3 2025: 65,668; down 76 %; data offering down 91 % | 10-Q Q3 2026 |
| Shutterstock, unsatisfied performance obligations, primarily data deals | 2026-09-30 | USD 33.2 million | expected over five years | 10-Q Q3 2026 |
| Getty Images, Other revenue share (includes data access and licensing) | 2023 / 2024 / 2025 | 1.9 % / 4.3 % / 5.6 % | Other also includes music licensing, digital asset management, distribution and print | 10-K FY2025 |
| Getty Images, royalties paid to content contributors | 2025 | over USD 220 million | all products, not AI alone | 10-K FY2025 |
DD&S is a revenue category, not a segment, and the Q3 2026 figure is one quarter. Shutterstock's 10-Q for the quarter ended 2026-09-30 also records USD 30.5 million of equity instruments received as revenue consideration, so some data income is not cash. Other companies' totals sit in AI data licensing revenue in company filings.
Lawsuits, a Terminated Merger and a Chapter 11: What Contributors Track
Getty Images licenses for display while litigating against Stability AI in the UK and the US, and in 2026 its merger with Shutterstock was terminated and it filed a prepackaged chapter 11. Getty Images (US) Inc, iStockphoto LP and Thomas M. Barwick, Inc are among the claimants in the UK claim IL-2023-000007, issued 2023-01-16, and Getty's own estimate of the photographers and contributors involved is "in excess of 50,000" ([2025] EWHC 2863 (Ch), para 59).
The High Court dismissed secondary infringement and found trade mark infringement in part on 2025-11-04. The Court of Appeal case CA-2026-000007 was part heard on 2026-11-11 with judgment reserved on 2026-11-12; there is no judgment as of 2026-11-17.
In the US, the N.D. Cal. case 3:25-cv-06891-TLT, filed 2025-08-14, is stayed until 2027-01-13 by Dkt. 81 of 2026-10-30 (as of 2026-11-16). The UK claims are tracked under Getty Images v. Stability AI in the UK courts.
The Getty and Shutterstock merger agreement of 2025-01-06 was terminated on 2026-07-07, according to Shutterstock's 10-Q for Q3 2026; Getty's 10-Q for Q2 2026 places this after the UK Competition and Markets Authority required divestment of Shutterstock's editorial business (Final Report of 2026-05-15).
On 2026-10-21 Getty Images Holdings, Inc. and affiliates filed chapter 11 petitions in the Bankruptcy Court for the District of Delaware (case 26-11643 (JKS)) under a prepackaged plan.
Its 8-K states that general unsecured claims are unimpaired and that contributors are paid in the ordinary course under first-day motions; solicitation was pending as of the filing.
Benefits, Risks, Downsides and Alternatives for Libraries and Contributors, Including Not Licensing
Licensing a stock collection to AI brings a new revenue line and contributor payments, but income has proved uneven, contributors see little of the terms, and a library or contributor may decide not to license.
Benefits
- Shutterstock's DD&S revenue rose from USD 137,323 thousand in FY2023 to USD 203,264 thousand in FY2025 (10-K for FY2025, filed 2026-03-16).
- Libraries state that contributors are compensated for AI use.
- Developers that train only on licensed data, such as Bria by its own statement, create demand for collections.
Risks
- Income is uneven: Shutterstock's DD&S revenue fell 76 % in Q3 2026 against Q3 2025, and part of the consideration is equity.
- No per-contributor opt-out is described, so a contributor's work may enter datasets without a choice.
- The contributor agreements' AI grant is not in the record, so what a library may license is not confirmed.
- A library in restructuring, as Getty is in chapter 11, may change terms.
Downsides
- No formula or rate is published by Shutterstock or Getty.
- Wirestock does not describe its license terms.
- Display deals do not state whether training is allowed.
Alternatives
- Display-only licensing, as in Getty's OpenAI and Perplexity deals.
- A commission route per license, as Wirestock states.
- Reserving rights and litigating, as the claimants in the Getty case have done.
- Not licensing at all, covered under alternatives to selling data to AI labs, including not selling.
Licensing, display-only licensing and not licensing are each valid outcomes.
Steps From Contributor Agreements to a Signed Library License
A library reaches a signed AI license in six steps, and the first one decides how much of the collection it may offer.
- Check that the contributor agreements grant AI and dataset licensing, and list exclusive contributors (Getty states over 83,000 exclusive among more than 600,000, 10-K FY2025).
- Separate works that have model and property releases from editorial content, and staff work made for hire under 17 U.S.C. 201(b) from contributor uploads.
- Clean captions and metadata.
- Decide whether to license for training, display or both.
- Set and publish the contributor compensation and reporting rule.
- Collect written terms from more than one developer and compare grant, term, deletion and payment form.
The same questions are asked of any holder in a rights audit.
Where Stock Libraries Meet Image and Video Sellers, Creators and Copyright Law
Stock libraries are one seller type among companies, publishers and creators, and the individual photographer or videographer selling outside a library is covered on the creator and artist pages.
A comparison of every type of seller sits in one place, with creators selling content for AI training and visual artists and illustrators on their own pages.
Footage sits under selling video for AI training, and the legal frame under copyright and AI training across jurisdictions. This page is information, not legal advice.
- Exclusive vs non-exclusive data licenses
- Voice and likeness rights in AI training data
- EU text and data mining exception
- Machine-readable TDM reservations
- Production studios licensing footage and images
Library, Deal and Case Pages Named on This Page
Each library, buyer, deal and case above has its own page with parties, dates and sources.
- Bria AI
- Getty Images v. Stability AI in the US
- OpenAI data buyer profile
- PIXTA AI
- Shutterstock's AI licensing deals
- Stability AI data buyer profile
Frequently Asked Questions
Do stock libraries pay contributors when their work trains AI?
Shutterstock and Getty Images state that they do, through a contributor fund and recurring compensation, and Wirestock pays a commission per license; none publishes a formula. Shutterstock and Wirestock were checked on 2026-11-09, and Getty's statement is from its 10-K filed 2026-03-16.
How much does a contributor earn from AI dataset licensing?
No library publishes a per-image or per-contributor amount for AI use, and Getty's USD 220 million in 2025 royalties covers every product, not AI alone. Adobe's bonus terms could not be read on 2026-11-09. Program rates and totals are compared under reported payouts to data sellers.
Does a stock library need contributor consent for each AI license?
That depends on the contributor agreement, which the record does not hold; Shutterstock describes Fund inclusion as automatic rather than chosen per dataset. The agreement text should be read for the AI grant before any answer is assumed.