Getty Images and the end of an era

For decades, Getty Images represented one of the pillars of global photographic licensing. A vast archive, an aggressive policy in defending rights, a comprehensive monitoring system for unauthorised images: Getty built its business model around the conviction that visual content had a precise value and that this value should be defended with legal instruments when necessary.

In 2023, the same Getty Images launched Generative AI by Getty Images^®, an image generation tool powered by artificial intelligence and trained exclusively on its own archive — content owned or licensed under explicit commercial agreements. In January 2024, the strategy was extended to the group’s more accessible segment with the launch of Generative AI by iStock^®, powered by NVIDIA Picasso^® and also trained on Getty Images’ creative libraries, with a standard legal coverage of 10,000 dollars for each generated asset. A few months later, it acquired Shutterstock^® in a deal worth 3.7 billion dollars, consolidating two of the world’s leading photographic archives into a single group.

The signal is unequivocal, and worth reading carefully.

From content custodian to Generative Capability provider

The traditional business model of photographic agencies is straightforward: someone produces content, the agency indexes it, distributes it and manages its rights at scale. The buyer pays to access that specific content, under defined conditions of use.

With AI generation, this model undergoes a structural transformation. The content does not exist before the user requests it: it is produced at the moment of the query, based on the parameters entered. There is no archive to browse, no photographer who shot that specific image, no file changing hands.

What changes is not only the technology, but the nature of the product. Getty no longer sells images: it sells the capability to generate images within an environment offering specific legal guarantees.

It is a paradigm shift we have already seen in other sectors: from product to service, from ownership to access, from the transfer of a good to the granting of a capability.

What changes is not only the technology, but the nature of the product. Getty no longer sells images: it sells the capability to generate images within an environment offering specific legal guarantees.

The differentiating value: Indemnification

Getty’s move is not solely technological. It is a response to a specific legal problem affecting most generative AI tools available on the market: uncertainty over training data.

The leading generative AI models have been trained on vast datasets of images collected from the web, often without explicit licences or compensation to the authors. This has generated significant litigation — Getty itself initiated legal proceedings against several AI model producers for unauthorised use of its archive.

Getty’s generative platform positions itself in the opposite direction: since the model is trained exclusively on content for which the company holds the rights, Getty can offer users a commercial indemnification guarantee. If a company is sued for the use of an image generated through the platform, Getty assumes legal responsibility.

This guarantee is not an incidental detail. For a company using visual content in advertising campaigns, packaging or institutional communication, legal certainty has a real and measurable economic value. The risk of a rights dispute — with the legal costs, reputational damage and potential penalties involved — is a risk that the legal departments of large corporations assess with growing attention.

This guarantee is not an incidental detail. For a company using visual content in advertising campaigns, packaging or institutional communication, legal certainty has a real and measurable economic value.

Collecting societies and the Remuneration Question

There is one aspect that the transition to generative AI leaves unresolved, and which the coming years will likely bring to the centre of debate: the remuneration of photographers and authors whose work contributed to training the models.

Getty has announced a compensation system for its contributors based on the use of images in the training process. The operational details of this system are still evolving, and the photographic industry watches with interest — and a degree of scepticism — how it will be concretely implemented.

The issue is not marginal. If the value of a generative model depends on the quality and variety of its training dataset, those who produced that dataset are entitled to participate in the value generated.

The issue is not marginal. If the value of a generative model depends on the quality and variety of its training dataset, those who produced that dataset are entitled to participate in the value generated. How this participation will be structured — through collecting societies, direct agreements, micropayment systems or alternative models — is one of the most relevant open questions for the entire creative industry.

What this evolution signals to the market

Getty’s move is not isolated. It represents a broader trend: the institutions that have historically managed and defended content rights are repositioning themselves as AI-enabled infrastructure, rather than opposing the technology or limiting themselves to fighting it in court.

For companies that purchase visual content, this opens new scenarios.

For companies that purchase visual content, this opens new scenarios. The relevant distinction will no longer be between traditional photography and generated imagery, but between content with documented legal guarantees and content without. The provenance of content — who produced it, with what data, with what rights — becomes a quality parameter as important as aesthetics or relevance.

For those working in communications, this means that competence in rights management — already essential in the traditional stock world — becomes even more critical in a context where the boundaries between original production, archival material and AI generation progressively overlap.

Getty Images spent thirty years building one of the world’s most important photographic archives. It used that same archive to train an AI model and place alongside its image sales a new product: the certainty of being able to use them. It is a shift in model that says a great deal about where the visual content market is heading — and what competences will be needed to navigate it.