The USPTO and the USCO Must Resolve Their Disparate Approaches to AI Inventorship and Copyrightability

The President’s recent Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence instructs the U.S. Patent and Trademark Office (USPTO) Director and Copyright Registrar to collaboratively issue recommendations to the President on further actions for advancing AI innovation through intellectual property, particularly with respect to AI inventorship and AI authorship. But the two offices currently regard AI differently in terms of assessing the creative and conceiving capabilities of machines, which poses a potential contradiction in how intellectual property law treats AI.

On the issue of AI inventorship, the question rests on whether AI contributed to the conception of an invention. The essence of the USPTO’s and the Federal Circuit’s decisions in Thaler v. Vidal is that conception is not presumed. As it stands, the USPTO definitely does not see machines as being capable of conception. However, the U.S. Copyright Office’s (USCO’s) position on AI authorship seemingly conflicts with the USPTO’s position, as the USCO adopts a de minimis test that automatically precludes copyrightability to any portion of work in which the AI’s contribution is determined to be more than de minimis.

Thus, the USCO’s de minimis test actually presumes creativity by the AI and fundamentally contradicts with the USPTO’s assessment of AI inventorship, which sees that AI is not capable of conception. If AI is not capable of conceiving by the USPTO, why is AI automatically presumed to be creative by the USCO? Similarly, if AI is capable of being creative before the USCO, then why is AI regarded as not capable of conceiving by the USPTO? How can the USCO’s de minimis test for precluding copyrightability exist concurrently with the USPTO’s position that AI is not capable of conceiving an invention? Before advising the President, the USPTO and USCO should not adopt contradictory approaches to AI’s contribution to the world of innovation and the useful arts.

The USPTO Does not Presume AI is Capable of Conception

The majority view by patent practitioners is that AI is not capable of conception that is “the formation in the mind of the inventor, of a definite and permanent idea of the complete and operative invention, as it is hereafter to be applied in practice.” Burroughs Wellcome Co. v. Barr Labs., Inc., 40 F.3d 1223, 1228 (Fed. Cir. 1994).

For example, in Corey Salsberg’s testimony to the Senate Judiciary Committee on AI inventorship, he noted that, despite generating 282 novel candidate molecules by Novartis’s AI-enabled research platform, JAEGER, only two were identified to have potential for further drug development. “JAEGER neither identified a problem, nor considered how to address it without prompting from human scientists and modeling from human-made precedents. Nor did it appreciate the properties or utility of its outputs, which had to be further analyzed, synthesized, developed and tested by humans before the results were realized,” Salsberg testified.

In the case of DABUS, the USPTO did not even examine Dr. Thaler’s invention, as his application was deemed incomplete for failing to list an inventor as a machine and was automatically determined to be incapable of conceiving under the current laws. Generative AI in its current form is widely regarded to be incapable of recognition and appreciation of the invention, which is required for inventorship.

The De Minimis Test Adopted by the Copyright Guidance Presumes AI Creativity

Under the USCO’s current rules, AI-generated content that is more than de minimis must be disclaimed. The USCO’s AI Guidance cites Feist v. Rural Telephone, which established the de minimis test for human authorship. “As a constitutional matter,” the Feist Court noted, “copyright protects only those constituent elements of a work that possess more than a de minimis quantum of creativity.”

The Guidance adopts the de minimis standard, citing Feist, as if this test as applied to AI authorship is based on historical precedent. However, the de minimis test adopted in Feist was applied as a standard for a finding of copyrightability while the de minimis standard now used for AI is a standard for precluding copyrightability.

In applying the test, the USCO makes a determination on whether the work would have been copyrightable had it been executed by a human. If yes, then copyrightability is precluded because the Copyright Act only permits human ownership of copyrights. But by viewing AI’s work as if it were created by a human, the USCO essentially assumes that the AI is being creative.

Take the case of Joseph Allen’s work in “Théâtre D’opéra Spatial”.  Here, the Board found that the artistic work contains more than a de minimis amount of content generated by AI and this content must therefore be disclaimed in an application for registration. Allen’s creativity clearly also met the de minimis standard, yet the AI-generated portion must be disclaimed. Artistic images created by integrating both AI and human input like Allen’s suffers under this test because having to disclaim the AI’s contribution so diminishes the work such that whatever remaining portion subsequently copyrighted fails to cover the artistic value of the integrated piece.

Historically, creativity was found based on finding of a work’s originality. The Copyright Act protects “original works of authorship fixed in any tangible medium of expression, now known or later developed, from which they can be perceived, reproduced, or otherwise communicated, either directly or with the aid of a machine or device.” 17 U.S.C. § 102(a). Copyright law only protects “the fruits of intellectual labor” that “are founded in the creative powers of the mind.” Trade-Mark Cases, 100 U.S. 82, 94 (1879). Associating creativity with originality makes sense when it comes to the human mind – one can reasonably assume that original works were birthed in the creative powers of the human mind.

But should that same assumption really be applied in the case of AI-generated works? Just because AI generated an original work, should the AI be deemed to be creative? If that is the case, then the corollary test in patent inventorship should allow for patentability of novel outputs generated by machines such as DABUS and JAEGER and view these machines as capable of conceiving. Or instead, should both patent law and copyright law consistently see AI as simply executing a series of computer implemented steps based on an algorithm?

The Issue of AI authorship arguably requires a much more nuanced approach when it comes to different levels of AI contribution in literary and visual works, which is a difficult matter. Copyright protection obviously should not extend to works generated solely by AI. But the current law already accounts for this situation by precluding copyrightability in the cases in which there is no human author. The difficult cases are the ones like “Théâtre D’opéra Spatial” and “Zarya of the Dawn” created from both significant human and AI input. Rather than adopt a split-the-baby approach under the de minimis test, which requires disclaiming AI contribution, the USCO should consider adopting a “substantial contribution test” to assess the degree of human input and allow for copyright protection to extend to the entirety of the work when met.

Image Source: Deposit Photos
Author: yacobchuk1
Image ID: 162843412