In January 2019, the USPTO held an AI IP policy conference. The conference featured IP specialists from around the world and included panel discussions on patents, trade secrets, copyrights, trademarks, IP enforcement, global perspectives, and the economics of IP protection of AI.
On August 27, 2019, the USPTO issued a request for comments (RFC) on patenting AI inventions. The USPTO received 99 comments from a wide range of stakeholders, including individuals, associations, corporations, and foreign IP offices.
On October 30, 2019, the USPTO issued a second RFC related to the impact of AI on other IP policy areas, including copyrights, trademarks, database protections, and trade secret law. The USPTO received 98 comments from a wide range of stakeholders, including individuals, associations, and corporations.
Following the conclusion of the comment periods, a team of experts assembled from across the USPTO to examine the responses and generate the report. The report is divided into two parts. Part I focuses on the first RFC solicitation dedicated to patenting of artificial intelligence technologies and provides AI context, legal background, and public comment synthesis, as appropriate, for each of the patent RFC questions.
Part II follows a similar format for the second IP RFC solicitation dedicated to non-patent intellectual property protections for artificial intelligence technologies, such as trademark, copyright, and trade secret. Below you can find briefly answers on some most interesting questions. All other questions and supplementary information you can find in full report.
Should a work produced by an AI algorithm or process, without the involvement of a natural person contributing expression to the resulting work, qualify as a work of authorship protectable under U.S. copyright law? Why or why not?
Under current U.S. law, a work created without human involvement would not qualify for copyright protection. However, a work created by a human with the involvement of machines would qualify for copyright protection if other conditions are met. The Supreme Court has long recognized copyright protection for creative works, even when an author is assisted by a machine.
The U.S. Copyright Office, in its Compendium of Practices (Third Edition), has addressed the question of human contribution to creative works. It notes that:
The Copyright Act protects “original works of authorship.” 17 U.S.C. § 102(a). To qualify as a work of “authorship,” a work must be created by a human being. …. Works that do not satisfy this requirement are not copyrightable. …. Similarly, the Office will not register works produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author.
Accordingly, the U.S. Copyright Office will not grant a copyright registration unless the author is a human being. A draft update to the Compendium further specifies that works “produced by a machine or mere mechanical process that operates randomly or automatically without any creative input or intervention from a human author” will not be granted copyright registration.
The United States is a member of the Berne Convention, the leading multilateral agreement establishing the framework for international copyright protection, which has been incorporated in large part in the TRIPs Agreement and subsequent U.S. free trade agreements. The Berne Convention has been interpreted to require protection only for works that are original and created with human involvement.
Assuming involvement by a natural person is or should be required, what kind of involvement would or should be sufficient so that the work qualifies for copyright protection?
U.S. law requires a minimum threshold of human creativity to qualify for copyright protection. A work’s copyrightability depends on whether creative expression, contributed by someone who can reasonably be described as an author of the work, is evident in the resultant work.
To the extent an AI algorithm or process learns its function(s) by ingesting large volumes of copyrighted material, does the existing statutory language (e.g., the fair use doctrine) and related case law adequately address the legality of making such use? Should authors be recognized for this type of use of their works? If so, how?
Existing statutory and case law should adequately address the legality of machine “ingestion” in AI scenarios. Mass digitization and text and data mining (TDM), as relevant examples of other activities with copyright implications, may be considered copyright infringement or fair use, depending on the facts and circumstances at issue. Copyright law in its current form appears to be adaptable to new technologies and circumstances, including those raised by AI.
Copying substantial portions of expressive (copyrighted) works, even for non-expressive purposes implicates the reproduction right and, absent an applicable exception, is an act of copyright infringement. Depending on the copyrighted work and the activity taking place, it may or may not be eligible for an exception to the reproduction right.
Regardless of whether an “ingestion” use is determined to be an infringement or not in a given situation, there is a separate issue of whether authors of ingested works should be remunerated for these types of uses. Many publishers now include TDM terms in their contracts and expressly set a licensing fee for for-profit entities or permit licensing at no additional cost for researchers and public research organizations, while ensuring that the licensed content is machine-readable and searchable.
Advocates for authors have suggested that when copyrighted works are used as inputs into AI systems to train the AI to create works of authorship or engage in other activities that result in remuneration, the authors should be entitled to a share of the revenues generated by the AI. The recognition sought is not attribution but rather remuneration.
The ingestion of copyrighted works for purposes of machine learning will almost by definition involve the reproduction of entire works or substantial portions thereof. Accordingly, whether this constitutes copyright infringement will generally be determined by considering the applicability of the fair use doctrine, an exception set forth in section 107 of the Copyright Act, 17 U.S.C. § 107. Fair use is applied on a case-by-case basis, requiring courts to weigh several statutory factors, and is highly fact-dependent.
When AI algorithms or processes “learn” their functions by ingesting copyrighted works, reproductions of those works are made in the process as the works are digitized and/or “read” by the AI algorithms or processes. Some mass digitization scenarios may be a fair use, whereas others may be infringements.
Although mass digitization for purposes of machine learning (ML) “ingestion” processes — and large-scale ingestion of already-digitized works — has not yet been tested by the courts, some rights holders argue that AI trainers should be required to compensate the authors and rights holders whose copyrighted works their machines are ingesting as a simple matter of doing business.
Are current laws for assigning liability for copyright infringement adequate to address a situation in which an AI process creates a work that infringes a copyrighted work?
While an AI machine cannot currently own intellectual property rights, it may be able to infringe others’ rights. Federal copyright law sets forth a straightforward standard for copyright infringement: “Anyone who violates any of the exclusive rights of the copyright owner” is liable for copyright infringement.
If the AI’s owner takes sufficient action to cause the AI’s infringement — through programming, data inputs, or otherwise — the owner could directly or contributorily infringe. Alternatively, if AI becomes more autonomous, it is conceivable that an AI owner might be vicariously liable for the AI’s copyright infringement when the owner possesses the right and ability to supervise the infringing conduct and a financial interest in the infringement.
Are there other copyright issues that need to be addressed to promote the goals of copyright law in connection with the use of AI?
The term “AI” can comprise a range of meanings. For example, generative algorithms (i.e., algorithms that possess the ability to create data) are responsible for producing unique works of varying complexity.
These works can result from collaborative efforts between a human creator and an AI program, or they can result from an independent AI process or algorithm. Therefore, no bright-line rule about “AI and authorship” or “AI and copyrightability” can be made; rather, it depends on a human being’s role in tandem with AI in generating a creative output that is potentially copyrightable.