Skip to content

Company and program review

Getty Images and AI licensing for creative imagery

Compare Getty Images stock, generative image and dataset rights before licensing creative imagery for AI.

Commercial basis
Stock, generator and dataset terms differ
Review updated
Provider website
gettyimages.com ↗

The verdict

Separate stock licensing, generated image use and dataset rights before relying on an agreement.

What to check: Ask which collection and releases cover the proposed model use.

How Fiund reviews companies and programs

Review basis. Fiund reviewed the cited public materials. Fiund also operates an AI data licensing business. This review includes no paid engagement or product rating.

Getty Images has a documented use of licensed creative imagery for generative AI. NVIDIA described the Getty Images generative product as trained on Getty’s creative library. That product announcement gives a concrete training example. A buyer seeking its own dataset should still request the current scope and availability from Getty. Fiund also licenses AI training data. NVIDIA on Generative AI by Getty Images

Ask which of three rights you need

  1. Stock display or editorial use. A license to illustrate an article or campaign sets terms for that use.
  2. Use of an image generation product. The service can grant terms for using its outputs.
  3. Access to images for your own model. This needs a separate agreement describing assets, allowed model uses, downstream access and retention.

Buyers should request the third scope in writing. Confirm whether a proposed package is creative or editorial, whether it includes model and property releases for the intended use, and whether the supplier can provide a provenance record for each asset. Ask Getty directly about a quote and whether the required modalities are available. Reviewed product materials do not establish a self-serve training dataset or a universal price.

What a photographer should check

Find the agreement that covers the specific account and collection. Review who can grant training rights, whether content is subject to representation or exclusivity, and how compensation works. An announcement about Getty’s own product cannot determine the payment due to a particular creator. Preserve a list of the titles and release records you can verify. If your work has appeared in several catalogs, compare the contracts before offering the same material elsewhere.

An original collection-scope exercise

Imagine a buyer receives a proposed list containing creative studio images, editorial event photographs and several near-identical retouched versions. Create separate rows for source class, release coverage, original asset and processed variant. Ask which listed files the proposed training grant actually includes.

For a performance sample, count distinct subjects and scenes alongside asset totals. Ten revisions of one composition do not provide ten independent examples of a new setting. Keep related variants together when forming evaluation splits. The exercise explains a sampling question; it claims no measured performance for Getty’s catalog or generation product.

A photographer should reconcile the actual account agreement

Record where each selected work is represented and which party can authorize the proposed use. Identify client ownership, contributor representation, participant releases and any exclusive commitments. Ask the provider how an individual training-related payment would be reported and what data the contributor can use to reconcile it.

A company announcement about a licensed model demonstrates a particular training relationship. It does not provide the licence for your own model or determine compensation for a specific photographer. Compare the written source and transaction scope for the exact collection before treating the product as a substitute for a dataset agreement.

Sources and further reading

Related reviews and guides

← Creative media and publishing