Getty Images Launches an AI-powered Image Generator

Getty Images, a prominent provider of stock images, editorial photos, videos, and music, has introduced a generative AI art tool it touts as a “commercially safer” alternative to existing solutions. This new tool, known as Generative AI by Getty Images, leverages an AI model provided by Nvidia, a close technical partner of Getty. The AI model was trained on a portion of Getty’s extensive library, encompassing around 477 million stock assets. Similar to popular text-to-image platforms like OpenAI’s DALL-E 3 and Midjourney, Getty’s tool can generate images from text descriptions or prompts, such as “photo of a sandy tropical island filled with palm trees.”

Getty’s Generative AI: Licensing and Safeguards

Customers who use the Generative AI tool to create and download visuals will receive Getty’s standard royalty-free license, which includes indemnification against copyright lawsuits and the right to “perpetual, worldwide, nonexclusive” use across all media. However, the tool is not without limitations. Getty has implemented safeguards to prevent misuse, disinformation, or the replication of living artists’ styles. For instance, the tool won’t allow users to create images of public figures in specific contexts or mimic the artistic style of renowned artists. Additionally, all images generated by the tool are marked as AI-generated with a watermark.

Getty’s Revenue Sharing and Accessibility

Getty commits not to add content generated by the tool to its library for licensing by others but reserves the right to retrain its model using these images. Contributors whose works were used to train the AI model will be compensated. Getty plans to share revenues generated by the tool, including a per-file proportional share and a share based on traditional licensing revenue. This offering can be accessed on Getty’s website or integrated into various apps and websites via an API. In the future, customers will have the option to customize the tool with proprietary data to align with specific brand styles or design languages. Pricing will be separate from standard Getty Images subscriptions and will depend on prompt volume.

Getty’s Stance on Generative AI and Intellectual Property

Before launching its own tool, Getty was critical of generative AI products like Stable Diffusion, which had been trained on a subset of Getty’s image content library. Earlier in the year, Getty filed a lawsuit against AI startup Stability AI, alleging the unauthorized copying and processing of millions of images and metadata owned by Getty without compensation to contributors. Getty’s CEO, Craig Peters, likened the current legal landscape in generative AI to the early days of digital music, emphasizing the need to address intellectual property rights. The debate over content scraping practices and fair use doctrine is ongoing and complex.

A Broader Landscape of Ethical Generative AI

Getty is not alone in pursuing “safer” and more ethical approaches to generative AI in commercial contexts. AI startup Bria offers a generative AI art tool trained on content licensed from partners, ensuring contributors receive a portion of the revenue. Ascendant Art, a recently launched avatar creator, commits to paying royalties to artists who voluntarily contribute their artwork to train its models. Established players like Shutterstock and Adobe are also exploring compensation models for contributors in the context of generative AI, aiming to support creators and navigate the evolving landscape of AI-generated content.