It took U.S. District Judge Beryl Howell all of 15 pages to dispense with AI developer Stephen Thaler’s argument that the U.S. Copyright Office and the U.S. Supreme Court have been misreading copyright law and the Constitution for the last 139 years. In granting the Copyright Office’s motion for summary judgment against Thaler last week in his lawsuit challenging the Office’s repeated refusal to accept registration for an image Thaler maintained was “autonomously created” by a generative AI system, Judge Howell reaffirmed the principle that copyright protection extends only to works created by human authors. Just as every court has done uniformly since 1884 when the Supreme Court upheld the constitutionality of extending copyright protection to photographs in Burrow-Giles Lithographic Co. v. Sarony.
“Human authorship is a bedrock requirement of copyright,” Howell wrote in her memorandum opinion accompanying her order granting summary judgment.
Thaler’s lawyers have vowed to appeal, but the odds will be stacked against him. As Howell noted, Thaler “can point to no case in which a court has recognized copyright in a work originating with a nonhuman.” Nor is the Supreme Court likely to be interested in the case, having so recently denied Thaler’s petition for review of the refusal by the U.S. Patent & Trademark Office to grant a patent for a device he also claimed was “autonomously invented” by an AI.
Thaler initially submitted the AI-generated image, titled “A Recent Entrance to Paradise,” in June 2019. The USCO rejected the registration, as well as Thaler’s two subsequent petitions for reconsideration, on grounds that the work, according to Thaler’s own submission, lacked human authorship. He then filed suit in January 2023 claiming the Copyright Office exceeded its statutory authority in refusing to register the image.
While Thaler apparently hoped to raise fundamental constitutional questions concerning the meaning of “author” and “authorship” in the Copyright Act, for procedural reasons his lawsuit was actually brought under the Administrative Procedures Act (APA), which governs how federal agencies carry out their statutory authority and issue rulings.
As Howell wrote, “The only question properly presented… is whether the Register acted arbitrarily or
capriciously or otherwise in violation of the APA” in denying registration. “The Register did not err in denying the copyright registration application presented by plaintiff. United States copyright law protects only works of human creation.”
As a legal matter, then, the court’s holding concerns only a narrow procedural question of administrative law rather than the loftier issues Thaler sought to raise. That is, whether the USCO appropriately construed and applied the law as it currently stands.
“This case presents only the question of whether a work generated autonomously by a computer system is eligible for copyright,” Howell wrote. “In the absence of any human involvement in the creation of the work, the clear and straightforward answer is the one given by the Register: No.”
Where exactly that leaves the legal status of such works, however, is not a question this case could or did resolve. Some technology activists, such as Electronic Frontier Foundation co-founder Cory Doctorow, declared the ruling means everything created by an AI is in the public domain and proclaimed it a “huge labor win” for creative workers. If you assume that any work not covered by copyright is, ipso facto, in the public domain, that would be true. But that begs the question in this case, as it wasn’t, legally speaking, the question before the court.
Celebrating it as a huge win for creative workers could also prove premature. Given the ease and scale at which generative AI can already churn out new, good-enough content, having thousands, millions of freely available public domain works floating around does not sound like recipe for a well-functioning, suitably remunerative market for works created by humans.
It is a question that Congress likely will need ultimately to resolve: If works created by AI are not eligible for copyright protection, do they, can they, should they have any legal status at all?