Getty Images now has its own AI-powered image generator tool. The photography giant is working with Nvidia to perfect a generator that takes advantage of its impressive content library. In this way, unlike the competition, will not have copyright problems and therefore, it would be a safer option for brands.
Getty Images is embracing artificial intelligence, after taking Stability AI, developers of Stable Diffusion, to court earlier this year. In a lawsuit filed in February in a US court, Getty Images claimed that Stability AI had used more than 12 million photos from its collection without authorization to train its famous artificial intelligence image generation model. He accused him of a “blatant violation of intellectual property.”
The big difference about their new service is that it is “commercially viable” for business customers and “was not trained on the open Internet with stolen images,” Craig Peters, CEO of Getty Images, told AP. In practice, anyone who uses its tool and publishes an image for commercial purposes will be legally protected, the US-based agency promises.
“Generative AI” from Getty Images will be available in Nvidia’s library of AI generation models, called Picasso. The company also has available on its website a form to request a demo of the new product.
Getty Images will compensate creators for their artificial intelligence model
The premise of use is the same as the artificial intelligence platforms launched so far: the user can describe in text what image they want and Getty Images gives it to them. The agency promotes that it has a file of 80 million images, from more than 1,500 collections.
Generated images will not be added to the library for licensing by others. Text descriptions from users to create new images will be used to continue training the model. Getty will mark the images so that it is known that they were produced by artificial intelligence, a security measure that companies like Google have already adopted.
The agency insisted that it will offer “full compensation for commercial use” to its users, in case of copyright conflicts. Microsoft promised something similar, with its Copilot Copyright Commitment, a program that guarantees its customers that they can use its generative artificial intelligence services without worrying about possible legal problems.
Getty Images ensures that the original authors will receive payment for the use of their photographs or images in this new generator powered by artificial intelligence. The company will share “income with them over time, rather than paying a one-time fee or not paying it at all”Peters said. However, he has not explained how or how much.
Limit for problematic images
Regarding the price of the tool, Getty Images said that it will have a separate cost from the standard subscription. He explained that it will be based on immediate volume, without giving further details.
Peters clarified that the new model does not have the ability to produce “deepfake” images. The tool incorporated filters to avoid, for example, the use of faces of recognized people or brands. Those who have accessed the demo have already tested the generator and verified that it is impossible to put together an image similar to that of Pope Francis with a Balenciaga coat, a viral one created by Midjourney in March that caused a sensation on the Internet.
Getty Images also said that customers will eventually be able to add their own data to train their model. In this way, you will have the ability to generate images with your brand’s own style. This and other new features would be available before the end of the year.
Meanwhile, Getty Images will press the lawsuit against Stability AI, which was also accused by a group of artists. The agency maintains that it is entitled to compensation of up to $150,000 for each infringed work. Technically, it could add up to $1.8 trillion in compensation. Stability, for its part, will try to dismiss the case, but has not formally responded to the allegations.