Last Friday, a U.S. court ruled that a piece of art created by AI isn’t open to copyright protection, upholding earlier guidance issued by the U.S. Copyright Office in March. But just how much human creative input is required remains an open question. It’s also the one that matters for any content company, including studios, and the larger creative community now considering using generative AI to create new material and IP.
The explosive rise of generative AI has brought with it a host of important legal questions on copyright grounds. A number of lawsuits have been filed by artists and creators, including one by Sarah Silverman, alleging that gen AI apps such as OpenAI’s ChatGPT and DALL-E infringe their copyrighted works.
Another battleground has been at the U.S. Copyright Office, where the department has refused to issue copyright registrations to works that were solely generated by prompting generative AI apps, “because the ‘traditional elements of authorship’ are determined and executed by the technology — not the human user.”
Underlying this policy is the requirement that copyright can protect only material that is the product of human creativity. This requirement is rooted in the understanding that the term “author,” as used in both the Constitution and the U.S. Copyright Act, excludes non-humans. But until now, this precise issue has never been tested by the courts.
The implications of whether and when AI-produced content can be copyrighted are significant. Generative AI tools are being more widely used to create content — including perhaps the next great piece of music, literature or cartoon character — and if that content is not protectable by copyright, it becomes free for anyone to use, unable to be monetized or licensed. This is why many have been closely watching the 2022 case filed by Dr. Stephen Thaler, in which he challenged the U.S. Copyright Office’s refusal to register a piece of visual art entitled “A Recent Entrance to Paradise.”
The art was generated by Thaler’s own AI system called “Creativity Machine,” or DABUS — described as an “artificial intelligence capable of generating original pieces of visual art, akin to the output of a human artist.” Thaler’s original copyright application in 2018 had identified the “author” as the Creativity Machine, with Thaler claiming the work was “autonomously created by a computer algorithm running on a machine.”
Recently, Thaler and the Copyright Office each made motions arguing the court should award it a victory as a matter of law. The single legal question presented to the court was “whether a work generated autonomously by a computer falls under the protection of copyright law upon its creation.” The Copyright Office argued that it does not. Thaler disagreed.
Stephen Thaler’s AI-generated work “A Recent Entrance to Paradise”
On Aug. 18, 2023, Judge Beryl A. Howell of the District Court for the District of Columbia sided with the Copyright Office. In an opinion granting its motion for summary judgment, he found that “human authorship is an essential part of a valid copyright claim.” The court found that the concept of “authorship” and the word “author” as used in the Copyright Clause of the U.S. Constitution and the Copyright Act means human creation and explained that this reading is consistent with the history and purpose of copyright law itself to “encourage human individuals to engage in that creation, and thereby promote science and the useful arts.”
While recognizing that copyright law is designed to adapt with the times, the court noted that “human creativity is the sine qua non at the core of copyrightability, even as that human creativity is channeled through new tools or into new media” and that “Copyright has never stretched so far as to protect works generated by new forms of technology operating absent any guiding human hand.”
AI is not the first technology to pose this type of challenge. Over the course of history, copyright law has “proven malleable enough to cover works created with or involving technologies.” For example, when the camera was first invented, there was a serious question and serious litigation over whether a photographer using that tool to create an image was entitled to a copyright on the resulting image.
Back then, the Supreme Court held the answer to that question was yes, because there was sufficient human creative input that goes into creating the image. While noting that “[a] camera may generate only a ‘mechanical reproduction’ of a scene,” the Court relied on creative decisions by the photographer, such as arranging the subject and composition, arranging the lighting and shading and crafting the overall image, in finding the resulting image copyrightable.
In some ways, despite involving a new and complex technology in generative AI, the Thaler case was easier, because Thaler asserted there was no human involvement whatsoever and therefore bore no creative contribution from a human. (Though Thaler later attempted to change course, asserting in the litigation that he “provided instructions and directed his AI to create the Work,” the Court stuck him with the assertions made when he was first seeking to register the work that it was “autonomously created” with no human input.) Once the court agreed the Copyright Act requires human involvement, that was the end for Thaler.
However, the court also acknowledged that tougher cases are coming down the pike: “Undoubtedly, we are approaching new frontiers in copyright as artists put AI in their toolbox to be used in the generation of new visual and other artistic works. The increased attenuation of human creativity from the actual generation of the final work will prompt challenging questions regarding how much human input is necessary to qualify the user of an AI system as an “author” of a generated work … and more.”
The Copyright Office’s current guidance on AI likewise recognizes that works containing AI-generated material can, in certain cases, contain sufficient human authorship to support a copyright claim: “For example, a human may select or arrange AI-generated material in a sufficiently creative way that ‘the resulting work as a whole constitutes an original work of authorship.’ Or an artist may modify material originally generated by AI technology to such a degree that the modifications meet the standard for copyright protection.”
The harder cases yet to come are ones in which there is some creative contribution by the human in the way the human collaborates with the generative AI. But until we have rulings by the Copyright Office on such copyright applications and further guidance from the courts, the question of just how much creative contribution is required remains largely open.
What we know so far is that, as per the district court’s decision, Thaler (which still may be appealed), AI alone with no human contribution cannot produce a copyrightable output. In addition, the Copyright Office has maintained that a text prompt alone is not sufficiently creative on the part of the human prompter to confer copyright protection.
But depending on the facts, one can imagine viable challenges to even this general conclusion. As people explore and practice the new and evolving art of prompt engineering and create more detailed and layered prompts, one could argue the human contribution to sophisticated prompting is more creative than the human contribution to snapping a digital photo with an iPhone.
And what about situations involving AI where the human does more than contribute a text prompt? There are already AI apps on the market that allow more creative control over the output — for example, NVIDIA’s eDiff-I’s “paint with words” tool and Stanford’s ControlNet tool — than providing a simple prompt and leaving the rest to the AI. Will these more sophisticated AIs that provide humans with more creative input and control yield copyrightable content?
It will be important for businesses and content creators to understand the newly forming rules around copyrightability so they can maximize their ability to gain IP protection for content created using AI tools. While Thaler is a first guidepost, we can expect that both the Copyright Office and the courts will be developing additional guidance on when AI can be used as a tool to create copyrightable output and on specifically how much human input is required.
Michael Kasdan is a partner in the intellectual property group at Wiggin and Dana LLP and the co-chair of the firm’s Blockchain and Digital Assets Group. His practice ranges from licensing to litigation to advising clients on IP strategy, and he’s a frequent speaker on the intersection of IP law and AI and other emerging technologies. He is also an adjunct professor at New York University School of Law, where he teaches a course on IP licensing.