A federal judge has upheld a decision by the United States Copyright Office to reject a copyright application for an artwork created by artificial intelligence in a ruling that could have significant ramifications in the generative art space. The ruling, delivered by District Court Judge Beryl A. Howell, found that Stephen Thaler’s copyright registration request for a visual artwork “autonomously generated” by AI was not subject to copyright protections, as the piece did not meet the agency’s human authorship requirement.
Thaler, founder of the AI company Imagination Engines, first applied for a copyright patent in November 2018 on behalf of his “Creativity Machine” — the AI algorithm-powered system that he claims created the artwork “A Recent Entrance to Paradise” (2018). In his application, Thaler listed himself as the owner of the machine and sought to register the artwork as a “work for hire.” After his original application was denied in August 2019, Thaler contested that the Copyright Office’s human authorship requirement was “unconstitutional and unsupported by either statute or case law.” But upon re-evaluating Thaler’s application, the agency again denied his request in March 2020, reasoning that Thaler had “provided no evidence of sufficient creative input or intervention by a human author in the Work.” When yet another request was denied in February 2022, Thaler sued the government office and Shira Perlmutter, the register of copyrights and the director of the Copyright Office.
In the ruling last Friday, August 18, District Court Judge Howell argued that “human authorship is a bedrock requirement of copyright” and that copyright “has never stretched so far … as to protect works generated by new forms of technology operating absent any guiding human hand.” She cited several legal cases in support of her decision, including the 1884 Supreme Court case Burrow-Giles Lithographic Co. v. Sarony, which extended copyright protections to photography on the basis that the resulting images were “representatives of the original intellectual conceptions of the author,” even though issued from a mechanical device. Among other examples, Howell also cited Naruto v. Slater (2018), widely known as the “Monkey Selfie” case, in which the court ruled “all animals, since they are not human — [lack] statutory standing under the Copyright Act.”
Thaler has tried (unsuccessfully) to obtain patents on behalf of his “creativity machine” in other countries, including the United Kingdom, Australia, and the European Union, according to Reuters.
Shyamkrishna Balganesh, a Columbia University law professor specializing in intellectual property and copyright law, explained that the ruling was “fully expected” given the precedents set by “prior caselaw.” However, he pointed to the limitation contained in Footnote 1 of the opinion, which notes that this ruling is confined to Thaler’s assertion that the work was created “autonomously” by his AI machine.
He added that “most artists who use AI systems in their creation are likely to rely on and readily acknowledge the creative input and intervention by a human.”
“Those cases present a closer question and require asking whether the computer merely assisted the human, such that it still satisfies the human authorship requirement, or whether it was essentially done by the machine,” Balganesh said.
Copyright expert Jane Ginsburg, who co-authored the 2019 paper “Authors and Machines,” agreed with Balganesh’s statement and said that copyright applications for other works — like author Kristina Kashtanova’s 18-page comic book “Zarya of the Dawn” (2022), which featured Midjourney-generated images alongside human-authored text — are trickier to discern. After receiving copyright protections in September 2022, the Copyright Office reversed its decision in February, instead granting Kashtanova partial-copyright protections for the story’s text and the book’s arrangement but not the AI-generated images as they were “not the product of human authorship.” In April, Kashtanova reportedly filed an application with the agency for a new AI work that is trained on an original drawing.
“It comes down to the question of, ‘how much do you control what the machine does?’ Have you constrained the machine a lot by giving it a starting point?” Ginsburg explained.
“The Copyright Office certainly didn’t say that works created using AI can’t be works of authorship,” Ginsburg went on. “Using AI ‘as a tool,’ like Photoshop, doesn’t disqualify you as an author, but where’s the line? At what point do you lose sufficient control over the output?”
In an email statement, Thaler’s legal counsel Ryan Abbott of Brown Neri Smith & Khan told Hyperallergic that they strongly disagree with the District Court’s opinion and plan to appeal the decision. “In our view, the law is clear that copyright is intended to benefit the American public through the generation and dissemination of new works, and this is accomplished when protection is provided regardless of how a work is made,” Abbott said.
The Copyright Office said in a statement that it supports Howell’s decision and is providing no further comment at this time.
The recent federal ruling trails several other legal cases in the past year pertaining to generative-AI authorship and intellectual property rights, in the wake of emerging software like ChatGPT, DALL-E, and Lensa that many creatives fear threaten their job security. The federal decision also arrived the same day that the Writers Guild of America (WGA) resumed talks with the Alliance of Motion Picture and Television Producers (AMPTP) over contract negotiations.
The use of AI-generative technology has been a major point of contention throughout the ongoing strike, which began on May 2, as strikers continue to call for regulations on the use of the technology in creative projects, among several other demands. Daily talks between the WGA and the AMPTP are expected to continue this week, according to a message circulated among WGA members last week.