Getty Images CEO Craig Peters has a plan to defend photography from AI

Summary

Getty Images CEO Craig Peters discussed the company’s approach to generative AI, including an image-generation tool developed with Nvidia and Getty’s efforts to protect creators’ work and distinguish authentic photography from synthetic images.

Getty’s AI image-generation tool

• The tool is trained only on Getty Images creative content, which Peters said is permissioned. Getty says it will share revenue with contributing creators.
• It is designed to be commercially safe and should not generate third-party intellectual property, recognizable public figures, or deepfakes. Peters said this is because the model was trained without the open-internet information that would allow it to identify those people.
• Getty provides indemnification for customers who use the tool.
• The tool can generate high-resolution images and guides users through prompts. In a live demonstration, The Verge editor-in-chief Nilay Patel generated scenes involving a conference, a celebrity couple, and a car-company CEO. Patel found the results impressive, while noting some visible AI flaws.
• Getty may also develop custom models using third-party intellectual property when rights holders provide the necessary permissions.

Creator compensation and business use

Getty does not currently have a reliable way to attribute an individual generated image to specific training images at the pixel level. Instead, it plans to calculate creator rewards based on:

1. The proportion of the training set represented by a creator’s work.
2. How that work has performed in Getty’s licensing business over time.

Peters said the approach may change if fairer or more precise attribution methods become available.

The tool could help customers create images for campaigns, but Peters said it is too early to know how it will compare economically with licensing existing photography or hiring photographers. Getty’s pre-shot images can be faster to find and may offer greater authenticity, including real people, locations, and brands. Generating a suitable image can take time.

Copyright, responsibility, and creators

Getty’s position is that intellectual-property owners should be able to decide whether their work is used to train AI and should be compensated when it is. Peters rejected the argument that training should automatically be treated as fair use and said AI companies should bear some responsibility for what their systems produce.

Getty has sued Stability AI over the use of Getty images in training. Peters described the case as high-stakes. He said Getty is also speaking with other technology firms about possible solutions, though he was unsure whether industry agreements would be enough to avoid relying on courts.

Peters framed Getty’s tool as a way to help creators make more work, rather than replace them. He said exceptional photography makes people stop, think, feel, or engage, and argued that creative expertise involves building empathy, trust, and meaning into images.

Authenticity and misinformation

Peters warned that generative images could make it harder to tell what is authentic, particularly at scale and in the context of elections and news. He cited a statistic that more AI images had been produced in the prior year than images taken with cameras.

Getty participates in discussions about content-authenticity standards but is not convinced that metadata-based approaches are sufficient. Metadata can be stripped, and the burden may fall too heavily on creators of authentic content rather than on platforms and generative-AI providers.

Getty is discussing possible responses with organizations including the Associated Press and Agence France-Presse, but Peters said there are no perfect solutions yet.

Reviews, demonstrations, and guides

• Product demonstration and review: Nilay Patel tested Getty’s tool with several prompts and described the results as remarkable, while pointing out some imperfections.
• Guides or tutorials: None were presented.

Main speakers and sources

• Craig Peters — Getty Images CEO; discussed the product, creator compensation, copyright, and authenticity.
• Nilay Patel — The Verge editor-in-chief and interviewer; demonstrated the tool and questioned Peters about its uses and limitations.
• Audience members — Asked questions about copyright law, generating public figures, and what makes a photograph exceptional.