Getty Images has released its own AI photo-generation tool called Generative AI, which will be available to its commercial customers. The tool, paywalled on the Getty.com website, allows users to generate new images based on prompts. It is trained on hundreds of millions of Getty Images using Nvidia’s model architecture, Edify. Getty Images CEO Craig Peters stated that the partnership with Nvidia allowed them to have unlimited graphics processing units to train the tool effectively. Competing with rivals such as Shutterstock and Adobe, Getty’s entry into the AI photo pool raises questions about the ethical implications of training AI models on photographers’ images and how photographers will be compensated.
Getty’s AI photo generator stands out from others due to its legally cleared images used to train the models. Peters asserts that it is commercially clean, making Getty Images the only offering that is fully indemnified. This means that if a customer uses an AI-generated Getty image that resembles original artwork, they are covered by Getty’s royalty-free licensing agreements. The tool also grants customers perpetual, worldwide, nonexclusive use of the images, and new content generated by AI will not be added to Getty’s existing content libraries for others to use. Getty’s move into the AI photo generation space brings competition to Shutterstock, which partnered with OpenAI, and Adobe, which integrated its generative AI engine, Firefly, into Photoshop.
