The advent of image generators from text has put stock illustration banks at a crossroads. They must decide whether to embrace artificial intelligence (AI) or reject it.
While Getty has decided to denounce Stability AI, the company behind Stable Diffusion, Shutterstock has taken a different tack. The latter has introduced its own generative AI tool.
Shutterstock and OpenAI join forces
The New York image bank, which is one of the largest on the Internet, now allows its paying users to create images from text directly from the platform.
The cornerstone of this possibility is an agreement established between OpenAI and Shutterstock, which allows the latter to use the technology behind DALL·E to generate images from text.
However, Shutterstock’s generative tool differs from others in one fundamental way. It is trained with its own images instead of huge datasets that possibly contain works protected by copyright.
The company says that it has worked so that artists “receive adequate compensation” in case their work is used to generate new content Thus, it says that it will pay them through a taxpayer fund, although it does not provide more details in this regard. .
It should be noted that this stance in favor of artificial intelligence does not mean that Shutterstock will accept any type of content generated by AI. In fact, the company it still prevents users from posting images generated by other AI tools.
The firm says that only images generated with Shutterstock’s AI will have the possibility of being licensed. That is to say, it will not be possible to upload works made with third-party tools such as Stable Diffusion or DALL·E 2.
“Given the availability of various AI content generation models on the market, we are unable to verify the source of the model for most AI-generated content,” Shutterstock adds in its usage policy.
Images: Cover generated with DALL·E 2 (Javier Marquez)
In Xataka: The expectation with GPT-4 is gigantic. Its creator is clear about what will happen: it will be a disappointment
