Following Shutterstock and OpenAI’s agreement to train the DALL-E 2 artificial intelligence generator on the former’s stock image libraries, the imaging site is now rolling out its own AI toolkit, allowing users to create pictures just by using text prompts.
Most notably, pictures created by the feature will be “ready for licensing” immediately after they’ve been generated. As TechCrunch points out, this could be significant for users, considering the rather murky copyright laws involving AI technology.
Getty Images, one of the site’s main competitors, is currently battling against Stability AI over its art generator Stable Diffusion, which allegedly uses the web page’s images to train its database without permission from the company or the images’ rights holders.
Starting now, users in every country will be able to access Shutterstock’s new image generation platform, and make use of its text-to-image technology to turn prompts into “larger-than-life, ethically created visuals ready for licensing.”
Need to get creative quickly, but short on time? Generate high quality visuals in a matter of seconds! Click to learn more about Shutterstock and how we are harnessing Generative AI technology. https://t.co/YPLdyUnegepic.twitter.com/LhQu9nkdpH
“Our easy-to-use generative platform will transform the way people tell their stories—you no longer have to be a design expert or have access to a creative team to create exceptional work,” explains Paul Hennessy, Chief Executive Officer at Shutterstock.
Hennessy emphasizes the site’s tools are “built on an ethical approach,” and that it helps ensure the artists whose works are involved in the development of the algorithm models are “recognized and rewarded” accordingly.
In addition, the tool accepts simple prompts—a single word or short phrases—in over 20 languages, taking away the chore of users having to write long, complex descriptions, and enabling people from all over the globe to bring visions to life.
Shutterstock claims to be the first to support such a “responsible AI-generation model,” offering artists compensation. Plus, the feature comes with built-in mitigations to guard against biases that may appear in datasets to “fairly depict underrepresented groups.”