A Generative AI Art Tool Launched

Generative AI art tool

Getty, a leading provider of stock photographs, journalistic photos, videos, and music, launched a generative AI art tool that it claims is safer than competitors.

Getty Images’ Generative AI tool, powered by Nvidia’s AI model, was trained on a fraction of their large library (477 million assets) of stock content. Getty’s program creates photos from written descriptions or prompts like “photo of a sandy tropical island filled with palm trees.”

Getty said users who create and download visuals using the tool will receive its usual royalty-free license, which covers indemnification and “perpetual, worldwide, nonexclusive” use across all media.

Recommended AI News: Amazon and Anthropic Announce Strategic Collaboration to Advance Generative AI

Drawbacks

Getty’s content repository contains popular figures, but it says it has measures to prevent its generative tool from being exploited for deception or copying a living artist’s style. Customers cannot make photos of Joe Biden in front of the White House or Andy Warhol-style cats, according to The Verge, which got access to the program before its release. The tool watermarks all photographs as AI-generated.

Getty claims it won’t add tool-generated pictures to its content library for licensing (but r***************** to retrain its model using those photographs) and will compensate creators whose works are utilized to train the model. According to Getty, it will split tool revenues into per-file proportional and traditional license revenue shares.

Recommended AI News: Cloudera Signs Strategic Collaboration Agreement with AWS

Getty vocally criticized generative AI solutions

The tool may be enabled on Getty’s website or incorporated into apps and websites via an API, and users can soon configure it with private data to create photos that match a brand style or design language. Getty says prompt volume-based pricing will be different from a typical Getty Images membership.

Before launching its own tool, Getty vocally criticized generative AI solutions like Stable Diffusion, which was trained on a subset of their image archive. Stable AI, which created Stable Diffusion, was sued by Getty earlier this year for allegedly copying and processing millions of Getty-owned photographs and metadata without telling or compensating authors.

In the U.S., several generative AI tool developers, like Stability AI, claim fair use protection for content scraping. This situation is unlikely to be resolved soon.

Bria, an AI firm, licenses content from photographers, artists, media companies, and stock picture repositories to train a generative AI art tool. Meanwhile, new avatar developer Ascendant Art promises to compensate artists who volunteer to train its models royalties.

Not just startups. Getty Images rival Shutterstock pays creators whose work trains AI art models. Adobe said it’s designing a remuneration plan for Adobe Stock contributors to “monetize their talents” and profit from Firefly’s earnings.