Getty Images—one of the world’s largest stock image libraries and aggregators—has recently launched its in-house AI image generator that will allow users to create their own images using prompts. In what is a competitive move to challenge other image generators such as Midjourney and Dall-E, Getty Images seeks to provide a clear pathway to use such images through its platform. While AI-generated images have been around for quite some time now, commercial firms offering generative AI services to create images and other creative products are relatively recent. This has also been the case with Adobe Firefly, which has constituted the firm’s major generative AI push in recent times. These developments hint toward the normalization of generative AI in spaces such as stock photography and digital design.
The fact that Getty Images has declared its AI-image generation protocol and its creations to be commercially safe is an important development. This is all the more significant when one considers the murky debate surrounding generative AI and copyright as well as the numerous related disputes. Moreover, approaches like those of Getty Images’ AI generator also allow platforms and developers to profit from their creations through ethical and legally acceptable means. Getty Images’ AI image generator is powered by NVIDIA—a major name in the hardware industry, adding further credibility to the service and reassuring the fence sitters surrounding the quality of the service. The subsequent sections explore more of the AI image generator and its capabilities.
Generative AI by Getty Images: Key Features
Generative AI by Getty Images is supported by NVIDIA’s Picasso foundry.
Getty Images’ generative AI was built while positioning the potential for business as the key area of focus. The commercially safe AI image generator is built with NVIDIA’s capable infrastructure and has been trained on a gargantuan database that is sourced solely from the parent firm’s suite of offerings. Getty Images utilized the Edify model created by NVIDIA to ensure the AI image generator is trained using a capable artificial intelligence model. All AI images generated by Getty Images’ generator are watermarked to indicate synthetic creation as opposed to contributor assets. Generative AI by Getty Images has trained extensively on actual photographs along with illustrations and digital artwork. Interestingly, the firm has said that the images generated by its AI tool will not feature on the image libraries hosted by the platform and iStock.
The tool is responsive and creates believable and realistic images with considerable flair. It also has AI safety features built into its framework and avoids responding to any prompts that demand the AI to generate images of public figures and real individuals. The generative AI tools are currently available only on the platform’s website and will be priced using a dynamic model that relies on prompt volume. Subscriptions for the AI tool will be distinct from the regular stock image models. However, the exact prices for the generative AI tool are not known yet. Generative AI by Getty Images has the potential to become a prominent player in the post-AI stock image space, where competition is stiff and several companies are consistently battling it out for vast chunks of the market share.
Normalizing AI Images: Why Getty Images’ AI Debut is Significant
Getty Images still needs to work on some of the finer details surrounding its novel AI image generator.
Getty Images’ entry in the AI images market signals a unique approach toward generative AI media. The firm has taken an ethical approach toward popularizing as well as monetizing its AI venture. Moreover, Getty Images has also announced that contributors whose assets have been used to train the underlying model will be duly compensated. This was also ratified by Getty Images’ CEO Craig Peters who said, “We’re excited to launch a tool that harnesses the powers of generative AI to address our customers’ commercial needs while also respecting the intellectual property of creators.” The partnership with NVIDIA has also been significant, since the latter’s “Picasso” foundry for visual generative AI applications and the Edify architecture has been instrumental in creating a viable AI image generation protocol for Getty Images. The firm’s entry into the generative space is also interesting because it had launched litigations against StableLM’s parent company—Stability AI—for having used Getty Images’ offerings without the requisite permissions for training its models.
The legal challenges faced by the firm have certainly pushed the company toward launching its own generative AI services. However, what’s unclear is how Getty Images plans on sorting out the exact contributors and their assets that will be included in the generation of any given image on the platform. However, despite details like these still being blurred in the overall process, Getty Images has been forthcoming in setting a benchmark for other AI image providers to follow while also crediting legitimate contributors whose work will be used in AI training. Such approaches might answer copyright problems with other applications of language models such as those seen in AI writing, eventually leading to more transparent AI usage policies. Regardless, there remain concerns surrounding the rights of contributors to refuse to grant the platform permissions to utilize their work in training their AI models.
The Impact of AI-Generated Photos and Media
AI-generated images will grow increasingly prevalent in media.
The impact of AI-generated content in various domains remains a hotly debated matter that has numerous implications for the way humans broach information at large. While the commercialization of AI-generated photos and images was already prevalent even before the launch of mainstream generative AI protocols, widespread formalization of these mechanisms will lead to increased prevalence. This will necessitate AI watermarks as well as stringent usage policies to prevent misuse and fraudulent deployment of artificial intelligence for ulterior motives. Moreover, further research and development of responsible AI protocols will also be necessary to ensure AI-generated photographs are not used to mislead, demean, or attack any individual or group of people. Though the road ahead looks promising, the approach to AI-generated images must be measured and informed with the requisite legal statutes.
FAQs
1. What is the price for Getty Images’ AI image generator?
Presently, Getty Images has only revealed that users will have to subscribe to the AI-image generation protocol separately. No prices have been listed as yet.
2. Does Generative AI by Getty Images use contributor images?
Yes, Generative AI by Getty Images uses individual contributors’ images to train the AI model. However, the contributors will be identified and compensated for their assets and their usage during the training process.
3. Will AI-generated images be available on the Getty Images directory?
No, user-created AI images will not be available on the directory and shall remain restricted to the individual user.