Image library service Shutterstock has decided to fully embrace AI-generated art through a new partnership with OpenAI, makers of popular AI image tool DALL-E 2 (opens in new tab).

This new partnership will allow customers to enter a request and have an image matching that request generated instantly. Essentially, it integrates OpenAI's core API into Shutterstock's platform, including custom adjustments.

Customers will be able to use this tool "within the next few months." [The data we licensed from Shutterstock was critical to the training of 'DALL-E,'" says Sam Altman, CEO of OpenAI.

"We look forward to future collaborations as artificial intelligence becomes an integral part of an artist's creative workflow."

In return for their help with the original DALL-E training, Shutterstock contributors will receive compensation for their role in the development of the technology; Shutterstock also states in the tool's FAQ (opens in new tab) that "Each artists involved in the creation of each new piece of content."

Are these contributors unwittingly signing their own death warrants and being paid a sum of money to do so, or are they grossly overestimating the capabilities of the AI art generation tool?

I side with the side that there will always be a place for human-made art. I would like to think that most people would agree with that view. However, there is no denying that the quality and capabilities of AI art generation tools are rapidly improving. These tools have emerged thanks to increasingly brilliant algorithms and large sets of human-generated images.

Shutterstock's model, in which AI is paid for its contribution to the images it generates, sounds at least fairer than the more common alternative in which artists' images are scrapped without compensation for use in datasets. However, questions remain about copyright and ownership of the images.

However, there are questions about what value AI-generated images have versus artist-created images, and whether they are comparable.

Another area of debate concerns the copyright of AI-generated art. These tools are trained on large data sets of images, and how these data sets are acquired is a concern. One artist actually found his own private chart photos (open in new tab) in a popular AI training dataset.

Shutterstock's main competitor and image platform, Getty Images, has banned all AI art (opens in new tab) from its platform entirely, citing copyright issues.

Speaking to The Verge (opens in new tab), Getty Images CEO Craig Peters said of the decision: "It's about who owns the copyright to that material, the rights that were used to create that material, etc. There are claims that the copyright is owned by x, y, z, or a particular platform, but I don't think those questions have been answered.

"I think we are seeing some organizations, individuals, and companies doing some reckless things, and I don't think those questions have been answered. I think the problem is that these questions have not been resolved. In some cases, these questions are just thrown by the wayside. I think that is dangerous. I don't think it's responsible. I think it could be illegal."

Shutterstock also does not allow others to upload AI-generated images to the platform unless they are generated by the integrated OpenAI tool. However, that does not mean that they are fully protected from copyright concerns.

Shutterstock asks users not to generate and use AI art that may infringe on the intellectual property of others, such as using recognizable trademarks, in the service's FAQ. It also asks users not to use these images for commercial purposes, stating that the use of a celebrity's name increases the likelihood that the tool will actually portray that person using the images contained in the image set.

This has always been a controversial topic. Companies at the forefront of digital image copyright management, such as Shutterstock and Getty, are now facing a new fork in the road brought about by AI image generation, and both companies are taking somewhat different paths. Ultimately, there is no simple solution, but many challenges to these AI models will be heard for many more years to come.