Getty Images announced a new AI art generator⁸⁵ tool trained on its massive stock image library. The company positions it as a safer, more responsible option than rival generative AI tools.

What's going on here?

Getty's new AI tool creates images from text prompts while aiming to prevent misuse and compensate content creators.

What does this mean?

The generator is powered by Nvidia and trained on Getty's library of 477 million images. It produces images from text prompts like "tropical island with palm trees". Getty says it has safeguards to avoid generating inappropriate content of public figures or mimicking artists' styles. All outputs have a watermark labeling them as AI-generated. Customers get a standard Getty license to use the images commercially.

Getty says it won't add AI-generated images to its main library, but may retrain its model with them. It plans to share revenues from the tool with contributors whose works trained the AI. This aims to address legal issues and compensate creators, unlike some other generative AI tools scraping content without permission.

Why should I care?

Getty's move shows major companies embracing AI art responsibly. This provides a commercially safer option for brands to use AI creatively. It also pioneers creator compensation models, helping content producers benefit from AI. If successful, Getty could influence the emerging generative AI industry to adopt ethical practices. This would ease legal concerns and make AI art tools more usable for businesses.

Getty Images earlier sued Stability AI for using their data to train Stable Diffusion. In the meanwhile, ShutterStock and Adobe have had their time experimenting with “safe” AI image offerings and might have leaped ahead of Getty Images.