INTRODUCTION

Generative AI (GenAI) has entered the fast lane in the industry race to the finish to create the best machine learning model as the technology continues to advance rapidly. We’ve seen GenAI image models like DALL-E, Midjourney, Stable Diffusion, Craiyon, and Imagen enter the hot seat for various scandals concerning topics like stereotypes and racism, copyright infringement, and unethical practices in the use and spread of imagery not considered to be “fair use” or in the public domain. As artists continue to hesitantly create and publish their work in a largely digitized world, how will they be able to trust that their work won’t be stolen without their permission?

THE GETTY

“The Getty” is the shorthand for J. Paul Getty Museum, a nonprofit located in Los Angeles, California and is a leading art museum in the United States in terms of attendance and unique visitor experience. As described on their website, the mission statement of The Getty is: 

“The J. Paul Getty Museum seeks to inspire curiosity about, and enjoyment and understanding of, the visual arts by collecting, conserving, exhibiting and interpreting works of art of outstanding quality and historical importance.”

It is from this museum that the world of arts and tech achieved significant advances and opportunities as the organization developed its own new business models.

WHAT IS GETTY IMAGES

This tangent off of The Getty Museum is a for-profit that brands itself as Storytelling that moves you. It also has the world’s largest privately owned archive, which has caused it to gain great popularity over time. The website goes on further to describe itself as:

“-the world’s foremost visual experts—capturing, creating and preserving content to elevate visual communications everywhere. By identifying cultural shifts, spearheading trends and powering the creative economy, we fuel visual storytelling worldwide. As photojournalists, videographers, creators, art directors, creative researchers, data scientists and customer service professionals, we’ve made it our mission to help you tell stories which will move your brand, your customers, your readers and viewers, and help you create lasting connections. Home to every kind of visual—from the dawn of photography to 8K video, and everything in‑between—we have the exact image, clip or illustration you need to boost your small business, inspire action or drive your global enterprise forward. Whether it’s commercial or philanthropic, revenue‑generating or society‑changing, market‑disrupting or headline‑driving, our visuals deliver an impact you and your customers can feel. Three brands, brought together by one mission: Move the world. Together, Getty Images, iStock and Unsplash are distinctly positioned to encourage and enable creativity and communications across the full spectrum of the world’s growing creative community.”

Getty Images offers various tools labeled creative, editorial, video, collections, and insights. Getty Images also owns iStock and Unsplash, two copyright-free imagery resources. The company offers various pricing options for those who use its resources on a frequent basis, which can be seen in the image below. In addition to this already successful business model, Getty decided to expand alongside the other tech giants and dive into the world of AI generation.

Image: Getty Images Pricing Plans

Image Source: Screenshot by author

GETTY IMAGES CREATES IN-HOUSE GENERATIVE AI

Getty Images has become the newest company to join the AI Image Generator bandwagon, creating their own that is trained using NVIDIA's Picasso. NVIDIA’s Picasso is a machine learning model that was created with the intention of using solely images that are property of Getty Images, which makes this Image Generator unique from the rest. 

Craig Peters, CEO at Getty Images, stated, “We’re excited to launch a tool that harnesses the power of generative AI to address our customers’ commercial needs while respecting the intellectual property of creators.” It is widely known that the intellectual property of artists has been widely contested in the world of Generative AI lately, hence the need for defensive tools like Nightshade AI and Kin.Art or Google’s DeepMind watermarking tool, to name a few. 

Image: Getty Images Home Page

Image Source: Screenshot by author

Image: Getty Images Home Page Breakdown

Image Source: Screenshot by author

Image: Getty Images Home Page Worry-Free Business Model

Image Source: Screenshot by author

Getty Images is marketing this new tool with the slogan, Commercially safe. Impactful. Worry‑free. Although the tool is not currently available for a trial, the user has the option to request a demo to get in contact with the Getty Images Sales Team to discuss the various plans and pricing. Getty is also marketing the new tool as an end-to-end solution, one that provides the user with the image library of Getty Images and allows them to use creativity to elevate their vision, stability that scales, with a computing infrastructure that provides speed and stability, and lastly one that compensates creators, through monetary exchange for the ability to use their images in the machine learning model.

COMPARING GETTY’S AI GENERATOR TO OTHER COMPETITORS

Much of this comparison was taken directly from AI Business’ article AI Image-Generation Models and Tools: The Ultimate List:

• Getty AI

• Creator: Getty Images

• First Published: Not accessible publicly yet

• Current Version: Beta

• Quote: From Craig Peters via My Modern Met, “We’ve created a service that allows brands and marketers to safely embrace AI and stretch their creative possibilities while compensating creators for inclusion of their visuals in the underlying training sets.”

• Other: Users can download all of their creations and even license them, and there's uncapped indemnification that comes with each download. Uses content from Getty Images’ own media library

• DALL-E

• Creator: OpenAI – San Francisco-based AI research lab backed by Microsoft

• First published: January 2021

• Current version: DALL-E 3

• Quote: From NBC News: “OpenAI released the second version of its DALL·E image generator in April to rave reviews, but efforts to address societal biases in its output have illustrated systemic underlying problems with AI systems.”

• Other: Can produce high-quality images based on text-inputs, in a similar format to ChatGPT. More than 1.5 million active users. Generates over 2 million images daily. Users are given 50 free credits to use when signing up (in exchange for an email address or Google account) and then another 15 credits a month; all other credits must be purchased.

• Midjourney

• Creator: David Holz, co-founder of Leap Motion (Now UltraLeap)

• First published: July 2022

• Current version: Version 6

• Quote: From Analytics Vidhya, “Midjourney, a popular text-to-image generator, is facing significant criticism and backlash following the recent release of its V6 models. Users are expressing concerns over potential copyright infringement. They claim that the new update's capabilities may be reproducing images too similar to copyrighted art.”

• Other: Only accessible via a Discord account, and therefore is housed on an official Discord server. Users directly message the bot, or they can invite the Midjourney bot to a third-party server. The bot has the capability to generate four images based on the request which users can then beautify.

• Stable Diffusion

• Creators: Stability AI - Based in London, Stability markets and manages the model. CompVis LMU - Research group from the Ludwig Maximilian University of Munich (LMU Munich) that created the deep generative neural network powering the model. Runway - New York-based applied AI research company building next-gen creativity tools using generative AI. Runway is responsible for the underlying algorithm that powers Stable Diffusion. LAION - A German nonprofit that built Stable Diffusion’s underlying dataset

• First published: August 2022

• Current version: Stable Diffusion 2.0

• Quote: From Unite.AI, “Stable Diffusion currently has no mechanism for focusing attention on the face during a render in the same way that humans prioritize facial information. Though some developers in the Discord communities are considering methods to implement this kind of ‘enhanced attention', it is currently much easier to manually (and, eventually, automatically) enhance the face after the initial render has taken place.”

• Other: Uses deep learning for generation; possibility for both image-to-image generations through text prompts and standard text-to-image generation. Can run on consumer-grade hardware − requiring a GPU with just eight gigabytes of RAM, setting it apart from DALL-E and Midjourney, which require cloud services to run. Looking to expand capabilities to be applied to audio, language, video and 3D generation for both consumer and enterprise use cases.

• Craiyon

• Creator: Craiyon - What started out as an independent research project by Boris Dayma turned into a popular image generator

• First published: April 2022

• Current version: Original

• Quote: From Cowan, DeBaets, Abrahams & Sheppard LLP, “Yet there is another layer here because, as noted above, per Craiyon’s terms of use, it is not only taking a cut of user commercialization, but also simultaneously disclaiming responsibility for its users’ exploitation of content that may result in legal claims by third-party content owners.  While social media platforms like Tiktok contain similar liability-shifting provisions and indemnification requirements, Craiyon’s demurrer of legal responsibility seems more like a necessary insurance policy given that it is unclear whether Craiyon is eligible for DMCA § 512(c) safe harbor protection that would shield it from third-party claims based on user-generated content.”

• Other: Formerly DALL-E Mini, Craiyon is designed to be a lightweight version of text-to-image models on this list. Craiyon is a free-to-use tool for non-commercial purposes. To use the model for commercial use cases, paid subscriptions are available. Users on the premium tiers have access to shorter wait times for their generations. Craiyon also relies on ads to pay for its servers.

• Imagen

• Creator: Google

• First published: May 2022

• Current version: Imagen 2

• Quote: From TechCrunch, “LAION is known to contain problematic content including but not limited to private medical images, copyrighted artwork and photoshopped celebrity porn — which obviously isn’t the best look for Google.”

• Other: Text-to-image diffusion model that uses transformer language models to understand text, and hinges on the strength of diffusion models to generate images in high fidelity. Its DrawBench benchmark for text-to-image models allows Imagen to be compared with other methods including VQGAN+CLIP, latent diffusion models and DALL-E 2. Latest version can generate logos and text across many languages. Used a version of the public LAION dataset to train the model.

HOW WILL THIS IMPACT THE ART INDUSTRY?

As artists continue to create new works but hesitate with the publication of their art onto the Internet for fear of being victimized by AI stealing their intellectual property, Getty Images is creating a platform that celebrates artists, compensates them, and then allows users to utilize this art to then build a foundation for their creative expression and create new art from the GenAI machine learning process. There is certainly a long way to go to ensure the protection of artists, but tools like this, alongside the watermarking and AI defensive tools that were created to combat this problem, there might be hope.