US Court Rules Copyrights For Generative AI Work Can’t Be Assigned To The Machines
August 21, 2023 by Dave Haynes
Hat Tip to Jason Cremins of Signagelive for passing along a Linkedin post about a court decision involving generative AI and copyrights.
This is the commentary of a Dublin lawyer, regarding a case in a Washington, DC court, filed by plaintiff Stephen Thaler, who has an AI system called the Creativity Machine, which generated a visual he dubbed A Recent Entrance to Paradise and wanted the copyright assigned to the AI system.
BIG NEWS! It’s finally happened. We finally have a decision from the courts on the status of AI generated works and copyright. A US District Court has published its decision in Thaler v. The Register of Copyrights, reaffirming the longstanding principle that human authorship is a requisite for copyright protection. I would be surprised if Thaler doesn’t appeal this decision.
For what it’s worth, my personal opinion is that the Plaintiff, Stephen Thaler is wrong here. He is trying to have the AI registered as owner of the copyright. I don’t think that’s possible in any jurisdiction. It would have been interesting if Thaler had tried to register the copyright in his own name, having used the AI system, how ever, he is saying that the AI created the work autonomously.
Thaler, sought copyright registration for the work of visual art below (called A Recent Entrance to Paradise) generated by an AI system he designed, known as the “Creativity Machine.” This AI system purportedly created the work autonomously, without any human involvement. Thaler’s application was denied by the US Copyright Office on the basis that the work did not meet the requirement of human authorship, which has been a fundamental element for the granting of copyright under the US Copyright Act of 1976.
Thaler’s arguments in support of his claim, namely principles of common law property and the work-for-hire doctrine, were rejected by the Court. The Court held that these doctrines were inapplicable since they presuppose the existence of a valid copyrightable work. In this case, the work created by the AI system was determined to lack human authorship and thus was not eligible for copyright protection.
The Court’s decision reinforces the belief that copyright protection should serve as an incentive for human creativity and artistic expression. This underscores the historical and foundational principle that copyright law seeks to protect works originating from human intellectual and creative efforts. In light of the Thaler decision, the Court’s affirmation of human authorship as a fundamental requirement for copyright protection is likely to continue shaping the legal landscape as AI systems become more prevalent in the creative process.
We must determine how best to balance the interests of human authors and the potential benefits and challenges of AI in the field of creative expression.
This is the first legal decision of many, many more to come. These AI issues are so new that the legal cases in relation. To text and say mining and AI copyright infringement have not yet even had the chance to make their way through the courts.
Thaler says he will appeal.
The Verge has a copy of the decision embedded in a post, if you want to read it.
Who knows where all of this goes. Anyone who has tried generative AI tools has likely concluded it is super-interesting, but also super messy and flawed. Scenics like the image at issue in the court case are easy to do. Real life visuals involving people and things – and especially text – are harder.
I’ve been trying to spend some time learning Midjourney, and while I am getting better, the outputs tend to be super-weird. Like what I got from the prompt: “people putting in order from inside car at drive-thru burger restaurant.”
I like the one that has the car hood covered in sesame seeds and the grill having a meat patty, lettuce and melted cheese.
Thanks for the mention, Dave.
Undoubtedly, the battle over creative IP will rumble on as generative AI moves from text to images and on to video.
Adobe’s position with Firefly is that they only use their stock footage to train their AI rather than the broader Internet to avoid legal battles.
Albeit the legal argument as to who owns the IP; the person who creates the prompt or the AI that responds with the generated content is the key decision.