Shutterstock combines a vast royalty-free library of photos, vectors, video and music with built-in AI generation that routes to models from OpenAI, Google and Runway. It suits marketers, designers and agencies who need commercially licensed assets, and it also sells training datasets to AI teams.
Used this tool? Rate it
Used this tool? Rate it
Shutterstock is a licensing marketplace that sells the right to use creative material. That single sentence explains more about the product than any feature list, because almost every question a new user has — why does this cost what it costs, why can I not use this image that way, why does the AI generator behave differently from the search box — traces back to the fact that you are buying a licence, not a file.
The company has been operating this marketplace since 2003, and its shape today reflects two decades of accumulation followed by a sharp pivot. The traditional half is a stock library: photographers, videographers, illustrators and musicians upload work, Shutterstock reviews and indexes it, customers search and licence it, and the original creator receives a percentage of what the customer paid. The newer half is an artificial intelligence business built on top of that same library — both generation tools pointed at customers, and training datasets sold to the companies building AI models.
Understanding the relative weight of those two halves matters, because it is easy to read the marketing and conclude that Shutterstock has become an AI company. The financial reality is more balanced. In the second quarter of 2026, revenue from the Content product line was $165.7 million and represented 75% of total revenue, while Data, Distribution, and Services accounted for the remaining 25%. Traditional licensing is still the main business by a wide margin. The AI side is real, growing in strategic importance, and increasingly what the company leads with in its own messaging — but it is the second pillar, not the foundation.
The company's own description of itself has broadened accordingly. Shutterstock now positions itself as offering access to one of the world's largest collections of licensable assets alongside specialized training datasets, evaluation tools, and end-to-end partnerships for the full model training lifecycle, as well as advertising and distribution solutions, exclusive editorial content, and full-service studio production. That is a considerably wider remit than "stock photos", and it is worth taking seriously as a signal of where the business is being steered.
A reasonable question for anyone browsing an AI tools catalogue is why a stock photo agency belongs here at all. Three things justify it.
First, generation is built into the product rather than bolted on as a separate service. Customers on the current subscriptions receive AI generation credits as part of the plan, in the same interface where they search the library.
Second, the generation is model-agnostic in a way most AI image tools are not. Rather than betting on a single proprietary model, the Unlimited plans bundle monthly AI generation credits with access to leading AI models from OpenAI, Google, Runway and more. Shutterstock is acting as an aggregation and licensing layer over third-party models, not as a model developer competing on raw capability.
Third, and least visible to end users, the company sells the raw material that other AI systems are trained on. Shutterstock describes itself as an end-to-end AI model training partner that unifies data licensing, services, and long-term collaboration under a single provider, aiming to reduce operational complexity for teams bringing AI systems to market. The library is simultaneously a product for creatives and an input for machine learning.
Anyone researching this platform will encounter extensive coverage of a proposed merger with Getty Images, and it is important to know how that ended, because a great deal of the surviving commentary was written while the outcome was still uncertain.
It did not happen. The long-running merger with Getty Images is over: Getty delivered a termination notice in July 2026 and the UK Competition and Markets Authority closed its remaining work on the case. The obstacle was regulatory rather than commercial appetite. UK regulators concluded the combination would reduce choice for British media outlets and could raise prices, and conditioned approval on Shutterstock divesting its editorial business. Getty's board declined that remedy and walked away instead.
The consequences landed directly on the balance sheet. Shutterstock recorded a non-cash goodwill impairment charge of $173.7 million in the second quarter of 2026, which the company attributed to the decline in its fair value following the announcement of the terminated merger agreement. For a prospective customer this is mostly context rather than a practical concern — the platform continues to operate normally — but it explains why the company is currently pursuing cost reductions and why its strategy leans harder on data licensing than it might have otherwise.
The library is the centre of gravity. It spans photographs, illustrations, vector graphics, video footage, music tracks and sound effects, along with editorial content covering news, sport and entertainment, which is licensed under different and more restrictive terms than commercial creative material.
Scale figures are best read from the plan definitions rather than marketing claims, because the plans state what is actually included. Unlimited Images covers a curated collection of more than 83 million premium images, while Unlimited Plus widens the pool to more than 100 million premium creative assets that also include video, music and sound effects. Note the word "curated" — these figures describe the pools available under those specific subscriptions, not the total catalogue, and the distinction matters when you are evaluating whether a plan will actually contain the material you need.
Search is where most of the daily value is created or destroyed, because a library this large is only as useful as your ability to find the right item in it. The platform offers filtering across the dimensions you would expect — media type, orientation, colour, licence type, people attributes, release status — plus reverse image search that lets you find visually similar material from an image you already have.
Discoverability is also the most common practical frustration. With tens of millions of assets, generic queries return enormous, homogeneous result sets, and the skill of using this platform well is largely the skill of writing specific queries and layering filters rather than scrolling.
The generation tool sits alongside search rather than in a separate product. You describe what you want, spend credits from your plan allocation, and receive images you can licence and download under Shutterstock's terms.
The company's public framing of this tool emphasises provenance. Its AI generator tool was trained on datasets licensed from Shutterstock's own library, which the company presents as ensuring copyright protection and compensation for the artists whose intellectual property was used to develop the technology. Whether you find that framing fully satisfying is a separate question, examined in the limitations section, but it is a meaningfully different posture from models trained on indiscriminately scraped web data.
Beyond search and generation, the platform includes browser-based editing so that a licensed asset can be adapted without a round trip to separate software — resizing for different placements, background removal, colour and filter adjustment, and template-driven layout for social and marketing formats. These tools are aimed at speed for non-specialists rather than at replacing professional design software.
This is the part of the business that will not appear in a typical creative user's workflow at all, but it explains a quarter of company revenue and much of the strategic direction.
Shutterstock sells rights-cleared multimodal training data to organisations building AI models, positioning clear data provenance as the core value proposition — the argument being that a model trained on licensed material carries less legal risk than one trained on scraped content. Around that data sit services: curation, custom dataset construction, and evaluation workflows involving human reviewers who score outputs, compare model results and supply preference signals used to tune models.
Revenue here is lumpier than subscription income by nature. Data, Distribution, and Services revenue was $56.1 million in the second quarter of 2026, down 16% year on year, and the company notes that recognition in the data offering may vary quarter to quarter based on the delivery timing of metadata licences.
Shutterstock is a group of brands rather than a single site. The company acquired Envato, the operator of Envato Elements and its associated marketplaces, in a $245 million cash transaction, and that acquisition brought both a large subscriber base and a catalogue extending into templates, themes and code assets that sit well outside conventional stock media. Users may therefore encounter Shutterstock-owned properties without recognising the parent.
The dominant use case. Campaign work needs many assets quickly, across many placements, with licences that unambiguously permit commercial use. A subscription providing predictable monthly volume fits this pattern better than per-item purchasing, and the combination of library search plus generation credits covers both "find something real" and "make something that does not exist" within one workflow.
Agencies licence on behalf of clients, which raises questions that individual users never face: who holds the licence, what happens when the client relationship ends, and whether the permitted distribution scale covers the client's actual reach. The standard commercial licence covers a great deal of ordinary usage, but high-volume distribution, merchandise and resale generally require enhanced terms. Agencies should resolve this before delivery rather than after.
Teams producing continuous social output need volume and variety more than they need any single perfect asset. Video footage, music beds and sound effects under one subscription reduce the licensing overhead of assembling short-form content, and template-based editing lets non-designers produce format-correct output.
A large and undramatic use case. Corporate decks, training material, internal portals and reports all consume imagery, and the value here is less about creative excellence than about having a clean licence trail for material distributed inside a company or to partners.
Editorial content — news, sport, entertainment imagery — is licensed for use illustrating events and reporting, and specifically not for commercial endorsement. Confusing the two categories is one of the more consequential mistakes available on this platform, because using editorial imagery in an advertisement is a licence breach regardless of how the file was obtained.
Retailers use stock imagery for lifestyle context around products, seasonal campaigns and category pages. Generation is genuinely useful here for producing backgrounds and contextual scenes that would otherwise require a photoshoot, though depicting an actual product accurately remains a job for photography.
A distinct audience with distinct needs: teams that require training data with documented licensing rather than creative assets for a campaign. They engage with dataset licensing, custom content commissioning and evaluation services, and the relationship is a negotiated commercial arrangement rather than a self-service subscription.
The supply side is also a use case. Photographers, videographers, illustrators and musicians use the platform as a distribution channel, and the earnings model is examined in detail later in this document because it has changed substantially in the AI era.
Before comparing prices, establish how the material will be used. Internal presentation, paid advertising, product packaging, a resold template and a broadcast production sit at different points on the licensing spectrum, and picking a plan before answering this question is the most common way to end up with either an overspend or a licence that does not cover the real usage.
There are broadly three shapes: recurring subscriptions with a monthly asset allowance, on-demand packs of a fixed number of downloads, and unlimited-style plans providing high-volume access to a curated pool. Read the commitment terms carefully at this step — the differences between plans that bill monthly and plans that bill monthly but commit annually are exactly where users have historically been caught out.
Write specific queries. Combine subject with context, mood or composition, then narrow with filters for orientation, colour and media type. If you have a reference image, reverse image search is usually faster than describing it in words. Save promising results to collections as you go rather than relying on rediscovering them.
Individual assets carry their own terms. Look for editorial-only designations, model and property release status, and any restrictions noted on the asset page. An image without a model release cannot be used in ways that imply endorsement by the person depicted, and this constraint applies no matter what your subscription allows.
When search does not surface what you need, spend generation credits. Write descriptive prompts covering subject, setting, style and composition, and expect iteration — credits are consumed per generation attempt, so vague prompts are expensive in a way that is easy to underestimate.
Use the built-in editing tools for resizing, background removal and format adaptation. For anything beyond routine adjustment, download and finish in dedicated design software.
Retain download records and licence documentation. If usage is ever questioned — by a client, a platform or a rights holder — the licence record is the evidence, and reconstructing it years later from memory is not realistic.
Match the plan to your real monthly volume, not your peak month. Subscriptions reward consistent usage and punish sporadic usage. If you genuinely need assets twice a year, on-demand purchasing is usually cheaper than a subscription you forget to cancel.
Read the commitment terms before the price. The most expensive mistakes on this platform are not about the sticker price but about the exit terms, which is documented in the limitations section below.
Search in the language of the library, not your own head. Indexed metadata reflects how contributors tagged their work. Describing what is visually present usually outperforms describing the abstract concept you have in mind.
Build collections continuously. Saving candidates as you browse is far more efficient than searching the same territory repeatedly across a project.
Treat generation as a supplement, not a replacement. Generated imagery is excellent for backgrounds, abstract concepts and scenes that do not exist. It remains weaker for specific real places, recognisable people, accurate product depiction and anything requiring documentary truth.
Write prompts that specify composition. Subject alone produces generic output. Adding framing, lighting, perspective and mood substantially improves the hit rate and therefore the credit efficiency.
Verify releases before using imagery of people. Model release status determines whether an image can be used in advertising contexts. This is checkable in advance and expensive to discover afterwards.
Never use editorial content commercially. It is the single clearest bright line in stock licensing, and the fact that a file downloaded successfully is not evidence that your intended use is permitted.
Diversify sources for distinctive work. Widely licensed stock imagery appears in many places at once. For brand-defining visuals, commissioned photography or original design still outperforms anything licensed from a shared pool.
Set a calendar reminder before any renewal date. Given the platform's documented history with renewal disclosure, proactive diary management is a reasonable defensive habit.
Marketing teams producing continuous campaign output, who need volume, licence clarity and predictable cost more than they need bespoke creative.
Designers and agencies assembling client work, who need breadth across media types and enhanced licensing options for high-distribution deliverables.
Small businesses and solo operators without photography budgets, for whom licensed imagery is the practical alternative to commissioning work.
Content and social teams publishing at high frequency, where speed and variety dominate.
Publishers and newsrooms requiring editorial imagery with reliable provenance and appropriate rights.
AI teams needing training data with documented licensing and clear provenance, engaging commercially rather than through self-service.
Contributors — photographers, videographers, illustrators and musicians — using the marketplace as a distribution channel, though with earnings expectations calibrated to the realities described below.
Who it is less suited to: occasional users needing one or two images, who will find per-item purchasing or free-licence alternatives more economical; brands requiring genuinely exclusive visuals, since anything in a shared library can appear in a competitor's campaign; and anyone whose usage cannot be pinned down in advance, since licence scope must be chosen before rather than after the fact.
The primary interface is the web application, which handles search, licensing, generation, editing and account administration and requires no installation.
Mobile access is available through the web interface and dedicated applications, with the contributor-facing mobile application allowing creators to upload work, track earnings and review market insights away from a desktop.
Integration into design workflows is supported through plugins for widely used creative and productivity software, allowing search and placement of licensed assets without leaving the host application.
Programmatic access is available through a REST API for organisations embedding search and licensing into their own products or automating asset workflows at scale. Detailed endpoint documentation is published by the company, though it could not be independently retrieved during the preparation of this document. (To be verified)
The contributor platform is a separate environment from the customer-facing site, with its own portal, help centre and account settings.
Because Shutterstock operates several brands, some capabilities are delivered through sibling properties rather than the main site, which is worth knowing when comparing what a given subscription actually unlocks.
Shutterstock is a paid platform. There is no meaningful free tier for licensing — free trials and occasional free asset offers exist, but the working assumption should be that using this material commercially costs money.
Purchasing falls into recognisable shapes:
Subscriptions provide a monthly allowance of downloads. Some bill monthly with month-to-month flexibility, others bill monthly under an annual commitment, and this distinction is materially more important than the headline rate.
On-demand packs provide a fixed number of downloads intended for defined projects, usable within a validity window.
Unlimited plans provide high-volume access to curated pools. Unlimited Images covers more than 83 million premium images; Unlimited Plus extends to more than 100 million premium creative assets including video, music and sound effects. Both include monthly AI generation credits and access to third-party AI models.
Enterprise and data licensing agreements are negotiated individually, covering elevated indemnification, team management and, for AI customers, dataset licensing and evaluation services.
On specific prices, this document deliberately declines to quote figures. Shutterstock's pricing pages could not be retrieved through either direct access or a full browser during research, and prices vary by region, currency, promotion and plan generation. Third-party sites publish figures freely, but reproducing unverified numbers in a reference document creates a worse problem than omitting them. Check the official pricing pages directly. (To be verified)
Two structural points are verified and worth internalising:
The word unlimited is doing lighter lifting than it appears: downloads are explicitly subject to Shutterstock's Fair Usage Policy and system rate limits, and customers who need more volume are told to contact support. "Unlimited" means high-volume with guardrails, not literally without limit.
And the plan you sign up for may commit you for longer than the billing cadence suggests. Regulators alleged that the plan selection page frequently failed to clearly disclose that annual paid monthly plans renew each year and that leaving early triggers a cancellation fee. Read the commitment terms before entering payment details.
Adobe Stock integrates natively with Creative Cloud, which is decisive for teams already living in Adobe software, and offers its own generative tooling with commercial terms.
Getty Images and iStock compete directly, with Getty strongest in editorial, news and sport, and iStock addressing the more price-sensitive segment. Getty remains an independent competitor following the terminated merger.
Envato Elements offers unlimited-style subscription access with particular strength in templates, themes and design resources. It is owned by Shutterstock, so choosing it is not a move outside the group.
Freepik and Vecteezy are strong on vectors, illustrations and graphic resources, often at lower price points and with free tiers subject to attribution.
Canva bundles a substantial asset library with design tooling, which suits non-designers producing marketing material end to end.
Unsplash and Pexels provide free imagery under permissive licences. The trade-offs are limited depth in specialised subjects, no indemnification, and high reuse of popular images.
Dedicated AI generators such as Midjourney and comparable tools generally offer more control and higher ceilings for image generation specifically, but no stock library and no licensed-asset fallback.
Pond5 and Artgrid specialise in video and are worth comparing where footage is the primary need.
The honest framing: Shutterstock competes on breadth, licence infrastructure and the combination of library plus generation in one place. It does not compete on being the cheapest, nor on having the most capable generative model.
This section is longer than is typical, because the documented issues here are substantive and a reader deciding whether to subscribe deserves them stated plainly.
This is the most serious concern and it is not a matter of scattered anecdote. In May 2026 the US Federal Trade Commission announced that Shutterstock would pay $35 million to settle allegations that it charged consumers without informed consent and made subscriptions difficult to cancel.
The specific allegations are worth reading carefully. Regulators alleged that on-demand packs marketed as suited to one-time projects with no commitment did not adequately disclose that packs renewed automatically when the last download was used, and — until early 2024 — renewed automatically after a year. They further alleged that the plan selection page frequently failed to clearly disclose that annual paid monthly plans renew each year and that leaving early triggers a cancellation fee, with such details buried in hard-to-find fine print. And on the exit path: before 2024, consumers could not complete early cancellation online and were required to contact customer support by phone, chat, or email, all of which regulators characterised as complicated and time-consuming.
The settlement was paid, and it is visible in the company's financial reporting — the $35.0 million payment for the settlement of the FTC's civil investigative demand appears in the second quarter 2026 cash flow discussion. Under the order, Shutterstock is now prohibited from misrepresenting material subscription terms and required to disclose those terms, obtain express informed consent, and maintain simple cancellation mechanisms.
Two things follow. First, the practices at issue have been the subject of a binding order, so the forward-looking position should be better than the historical one. Second, the historical pattern is documented rather than alleged-and-unresolved, and prospective subscribers should read commitment and cancellation terms with corresponding care.
User sentiment reflects the same pattern. On Trustpilot the platform carries a 1.2 rating across roughly 2,762 reviews, with 93% of them at one star and the complaints clustering around subscription terms, automatic charges and cancellation. Review platforms attract dissatisfied users disproportionately and this figure should not be read as a satisfaction survey of the whole customer base — but the concentration of the complaints on exactly the issues regulators pursued makes the signal harder to dismiss than a generic low rating would be.
As noted above, unlimited plans operate under a fair usage policy and system rate limits. Users planning genuinely extreme download volumes should establish the practical ceiling before committing.
A standard licence does not cover everything. High-volume distribution, merchandise for resale, and use in templates or products intended for onward sale typically require enhanced licensing. Editorial content cannot be used commercially at all. Assets lacking model or property releases cannot be used in ways implying endorsement. None of this is unusual for the industry, but all of it is easy to breach accidentally when a file downloads without complaint.
Everything in a shared library is available to everyone else, including competitors. Popular images become recognisable through repetition. For brand-critical visuals this is a genuine strategic limitation rather than a minor annoyance.
AI generation consumes credits per attempt regardless of whether the output is usable, so poor prompt discipline translates directly into cost. Generated output remains weaker for real locations, identifiable people, accurate product depiction and precise text rendering. Publicly visible customer complaints about credits being consumed on unusable generations exist and are consistent with how credit-metered generation behaves generally.
The terminated Getty merger, the associated $173.7 million goodwill impairment, declining revenue in both product lines, and ongoing cost reduction are the current backdrop. This does not affect day-to-day usability, but customers making multi-year commitments may reasonably factor strategic uncertainty into that decision.
The ethical dimension of this platform deserves its own treatment, and the next section provides it.
Any honest assessment of an AI-era stock platform has to address what happens to the people who supplied the library, because their work is the input to both halves of the business.
This is not disputed or hidden. Shutterstock states that its AI generator tool was trained on datasets licensed from Shutterstock, presenting this as ensuring copyright protection and proper compensation for artists whose intellectual property was used to develop the technology.
The compensation mechanism is the Contributor Fund, and its structure matters. Earnings from datasets and downloads of AI-generated content are pooled in a collective fund and distributed periodically once that fund accumulates significant revenue, rather than settled per transaction. Individual shares are proportionate to the volume of a contributor's content and metadata included in the purchased datasets.
Read carefully, that design has clear implications. Payment is pooled rather than attributable to a specific use, periodic rather than continuous, and proportional to catalogue volume — which structurally favours contributors with large portfolios over those with small but high-quality ones. Whether that is fair is a judgement call; what it is not is opaque, since the company documents the mechanism publicly.
The company's headline figure on contributor payments is that it has paid contributors more than $1 billion since 2003. That is a vendor-reported figure covering all payment types across two decades, and it has not been independently audited. It should not be read as an indication of what any individual contributor earns.
This is the most practically important fact for creators weighing participation. Since January 2023 the contributor account settings have included an option to opt out of having your content included in future datasets, with separate controls subsequently added for image and video data licensing.
The limitation is in the word "future". An opt-out governs subsequent dataset inclusion; it does not retroactively remove content from datasets already licensed or from models already trained. Creators concerned about training use should understand that timing is decisive.
A detail that surprises people, and which is more principled than it first appears: Shutterstock will not allow AI-generated content to be submitted by contributors for licensing on the platform, a policy that separates the machine-generated supply from the human-supplied library.
The stated reasoning is twofold. Contributors must be able to prove intellectual property ownership of submitted work, and because generative models draw on the intellectual property of many artists, ownership of AI-generated content cannot be assigned to an individual. Separately, the company says it cannot verify the source model for most AI-generated content and therefore cannot ensure that all artists involved in generating a piece are compensated.
The result is a coherent, if striking, arrangement: the platform generates AI images itself and licenses them, but declines to accept AI images from its contributor community. The distinguishing factor is provenance — it can account for what its own tool was trained on and cannot account for what an arbitrary external model was trained on.
Contributor income from conventional licensing is a percentage of what the customer paid, and that percentage varies with the earnings level the contributor occupies at the time of purchase, up to a stated ceiling of around 40% of the price paid. Levels reset annually, which the company frames as ensuring contributors are rewarded for content performing well currently — the practical effect being that established contributors restart each year at lower rates and work back up.
The role has also broadened well beyond uploading files. The platform now presents three earning routes: selling multimedia content, powering AI training, and getting paid for creative expertise. In practice that means traditional licensing, commissioned content created specifically for AI and machine learning training, and task-based work such as annotation, curation, editing, review and scoring in service of AI development. Creators are being invited not just to supply training data but to participate in the labelling and evaluation labour that model development requires.
For customers, the practical question is whether generated images are safe to use commercially. Shutterstock's position is that AI-generated content produced through its tools is covered under its licences in the manner of other content on the platform, with indemnification available consistent with customer agreements. Because the detailed licence and indemnification pages could not be retrieved during research, buyers with meaningful legal exposure should read the current licence terms directly rather than relying on any summary. (To be verified)
Worth noting separately: copyright status for purely AI-generated images is unsettled in several jurisdictions independent of what any platform's terms say. A licence from Shutterstock governs your relationship with Shutterstock; it does not resolve whether an AI-generated image is capable of copyright protection under a given national law.
No, not for licensing. Browsing and searching are free, and the company periodically offers free trials or free asset selections, but downloading and using material commercially requires a paid subscription or purchase. Registration is free for contributors who want to sell work.
Generally yes under a standard licence, but with real boundaries. High-volume distribution, merchandise for resale and inclusion in products sold onward typically require enhanced licensing. Editorial content cannot be used commercially at all. Assets without model or property releases cannot be used in ways implying endorsement. Check the terms on the specific asset rather than assuming your plan covers every use.
Generated images are licensed to the customer who created them under Shutterstock's terms, in a manner comparable to other platform content. However, copyright protection for purely AI-generated work is legally unsettled in several jurisdictions regardless of platform terms. Organisations with significant exposure should review current licence documentation and take their own legal advice. (To be verified)
Yes, and the company documents this openly. Its generation tool was trained on datasets licensed from its own library, and it also licenses datasets to external AI developers. Contributors are compensated through the Contributor Fund, which pools earnings from datasets and AI-generated downloads and distributes them periodically in proportion to how much of a contributor's content appears in purchased datasets.
Yes. Since January 2023 contributors have been able to opt out of having their content included in future datasets through account settings, with separate image and video controls added later. The opt-out applies to future datasets and does not retroactively withdraw content from datasets already licensed or models already trained.
No. Shutterstock does not accept AI-generated content from contributors for licensing. The company's reasoning is that ownership of AI-generated content cannot be assigned to an individual, and that it cannot verify which model produced a given image and therefore cannot ensure the artists whose work informed it are compensated.
High volume with guardrails rather than literally without limit. Unlimited downloads are subject to Shutterstock's Fair Usage Policy and system rate limits, and customers needing more are directed to contact support. The plans also draw from curated pools — more than 83 million images on Unlimited Images and more than 100 million assets on Unlimited Plus — rather than from the entire catalogue.
It has been a documented problem. The FTC alleged that before 2024 consumers could not complete early cancellation online and had to contact support by phone, chat or email through a complicated and time-consuming process, and that annual paid monthly plans carried early cancellation fees disclosed only in fine print. A $35 million settlement in May 2026 now requires clear disclosure of material terms, express informed consent and simple cancellation mechanisms. Read commitment terms carefully before subscribing and note renewal dates.
No. The merger was terminated. Getty delivered a termination notice in July 2026 after UK regulators conditioned approval on Shutterstock divesting its editorial business — a remedy Getty's board declined. The two companies remain independent competitors, and Shutterstock recorded a $173.7 million non-cash goodwill impairment in the second quarter of 2026 following the termination.
Still mainly a stock library, though the AI side is strategically prominent. In the second quarter of 2026 Content revenue was $165.7 million, or 75% of total revenue, while Data, Distribution, and Services — which includes AI training data licensing — made up the remaining 25%. The AI business is the growth story and the direction of travel, but traditional licensing remains the larger share.
