Legal Questions, Lawsuits Abound as AI Art Generators Continue to Gain Steam

Artificial intelligence (“AI”) artwork generators garnered a notable amount of attention in 2022 and have continued to do so this year. One of the most widely referenced AI art generators, DALL-E – which was first introduced in January 2021 – is touted as capable of “creat[ing] original, realistic images, and art” based on a user-provided text description.

Copyright: thefashionlaw.com – “Legal Questions, Lawsuits Abound as AI Art Generators Continue to Gain Steam” Image: Valentino

In other words, if you type in a combination of words, DALL-E can render an image based on that input using machine learning and supporting software. Other image generators, such as Stable Diffusion, operate similarly, creating images that look like oil paintings, drawings, photos, etc. based on a user’s prompts, or music (based on genre, length, and mood inputs) with little – if any human – contribution. Unsurprisingly, the ability of these AI systems to generate artistic outputs presents legal questions, namely, in the copyright vein.

As AI takes on a larger role in creation, as reflected in AI-generated ad campaigns coming from brands like Valentino and spirits company Martini, “It is becoming increasingly important to determine whether there are intellectual property rights [at play when it comes] to substantially machine-generated works and, if so, who owns them,” Lewis Rice’s Kirk Damman, Benjamin Siders, Kathleen Markowski Petrillo, and Olivia Dixon stated in a note this summer. Primarily, there is the issue of copyright protection in the U.S., which has been reserved for works of human authorship, and thus, poses questions about whether images entirely created by AI could be deemed eligible. (Registration is significant, as it is a prerequisite for filing a lawsuit for copyright infringement.)

“Courts have in the past decided cases revolving around the ‘human authorship’ requirement of the Copyright Act (e.g., pictures taken by monkeys are not copyrightable because they are not created by a human),” according to Mesh IP Law’s Michael Eshaghian. But until recently, courts had “not been directly faced with the question of whether AI-generated works are copyrightable.” However, that is changing (and the law could be changed – by courts or Congress) thanks to the lawsuit that Stephen Thaler waged against the Shira Perlmutter and the U.S. Copyright Office in the wake of the U.S. Copyright Review Board confirming that copyright protection only extends to works of human authorship.

Beyond the threshold issue of registrability, other questions abound, according to Morrison & Foerster’s Heather Whitney and Tessa Schwartz, such as: (1) to what extent “can a human use AI tools in the creation of a work before that work will no longer be considered a work of a human author” and (2) “whether the outputs from certain deep learning AI systems can infringe copyrighted works.” In terms of AI-generated outputs, they assert that aside from violating the exclusive rights of the copyright owner (from unauthorized copying to the creation of derivative works), courts also require that any alleged infringement “result from the defendant’s ‘volitional conduct.’” This is likely to prompt debate over “whether infringement occurs where outputs of sophisticated deep learning AI systems are attenuated from anyone’s (i.e., a human’s) volitional act.” […]

Share this:

Related