July 11, 2020

Volume X, Number 193

July 10, 2020

Subscribe to Latest Legal News and Analysis

July 09, 2020

Subscribe to Latest Legal News and Analysis

July 08, 2020

Subscribe to Latest Legal News and Analysis

USPTO Confirms Inventorship as Limited to Natural Human Beings

A key question surrounding patents for artificial intelligence (AI) related technologies is can an AI agent (as in, not a human) be considered an inventor to a patent application. Last week, in a highly publicized case, the United States Patent & Trademark Office (USPTO) answered with a resounding “no” to an AI agent named “Device for Autonomous Bootstrapping of Unified Sentience” (“DABUS”) that “autonomously generated” the invention in question. The decision explained that an AI agent cannot meet certain statutory definitions for an inventor or the jurisprudential tests for determining inventorship.

In the USPTO’s view, the language of the Patent Act suggests an inventor must be a natural person because it uses terms such as “whoever,” “himself or herself,” and “individual” when describing an inventor. The Act also describes actions that can only be performed by a human being, for example, by directing an inventor to “execute an oath or declaration.” The decision further cites case law defining the contours of inventorship, which turns on the question of “conception.” Generally, the legal tests for conception require a person to conceive of and contribute to the inventive subject matter in the claims. According to the USPTO, these tests contemplate conception in the “human mind” and therefore only natural persons can satisfy the requirements for inventorship. 

The DABUS Application

The applicant, Stephen Thaler, in conjunction with the Artificial Inventor Project, filed a patent application for Devices and Methods for Attracting Enhanced Attention in various jurisdictions around the world, including the United States. The application was highly publicized because it listed DABUS, an AI agent, as the sole inventor. 

Unlike the inventors for every other patent application filed at the USPTO, DABUS is not a human being. Rather, it is an AI agent created by a team of researchers designed as a “creativity machine” to mimic the creative problem solving skills of a natural person. Thaler reasoned that since the invention disclosed in the patent application was generated entirely by DABUS, without any human involvement, DABUS was owed inventorship credit. Thaler thus listed DABUS as the sole inventor accordingly in the patent application’s filing documentation.

The application, however, was objected to at the pre-examination phase when compliance with various formality requirements is evaluated. Specifically, the USPTO issued a Notice to File Missing Parts rejecting the application as incomplete, stating that the documentation failed to identify an inventor as required. Thaler petitioned for an administrative review of the alleged deficiency, arguing that inventorship is not limited to natural persons and DABUS was properly identified as the inventor. The Office of Petitions disagreed, finding that an AI agent cannot be an inventor under US law, thus confirming the USPTO’s view on the overarching question—only natural persons can be inventors.

The Decision

The USPTO’s decision was based primarily on its interpretation of certain portions of the Patent Act (35 U.S.C. § 100 et seq.) pertaining to who is an “inventor” and what acts are performed by an inventor. According to the USPTO, these sections of the Patent Act indicate that an inventor must be a natural person: 

  • 35 U.S.C. § 100(f), (g) each define an “inventor” as an “individual.” 

  • 35 U.S.C. § 101 provides that “[w]hoever invents or discovers . . . may obtain a patent.” 

  • 35 U.S.C. § 115(a) requires “each individual who is the inventor” to execute an oath or declaration.

  • 35 U.S.C. § 115(b) requires the oath or declaration to “contain statements that . . . such individual believes himself or herself to be the original inventor.”

  • 35 U.S.C. § 115(h)(1) permits “[a]ny person making a statement” in an oath or declaration to withdraw, replace, or otherwise correct the statement.

In the USPTO’s view, these sections, individually and in combination, suggest that an inventor is contemplated statutorily as a natural person. Consequently, the decision on Thaler’s petition observed “interpreting ‘inventor’ broadly to encompass machines would contradict the plain reading of the patent statutes that refer to persons and individuals.” 

The decision also cites several Federal Circuit cases to support its interpretation, asserting that the court’s view is that “the patent laws require that an inventor be a natural person.” In Univ. of Utah v. Max-Planck-Gesellschaftzur Forderung der Wissenschaften E.V., the Federal Circuit held that a state cannot be an inventor. Furthermore, in Beech Aircraft Corp. v. EDO Corp., the court distinguished “inventorship” from corporate “ownership,” and expressly noted that “only natural persons can be ‘inventors.’” In both cases, the Federal Circuit visited the notion of “conception” as the “touchstone” test for inventorship, which has been described as a “mental act” and “the completion of the mental part of invention” for roughly a century. Specifically, conception “is the formation in the mind of the inventor, of a definite and permanent idea of the complete and operative invention.” In the USPTO’s view, even though the cases cited in its decision “are in the context of states and corporations,” they still frame conception and inventorship as a result of human thought and necessarily require an inventor to be a natural person. A machine, such as DABUS, therefore cannot meet the requirements for inventorship.

This decision effectively precludes AI agents from being listed as inventors in the US. Going forward, it will serve as administrative guidance to USPTO staff who evaluate whether applications are filed with all required documentation. Unless and until a Federal court holds that an AI agent can meet the statutory requirements to be an “inventor” or Congress legislatively redefines the definition of “inventor” to include an AI agent, only natural persons are eligible to be inventors. But this result is unlikely to change for the foreseeable future. Except for a handful of specialized advocacy groups and a few academics, the stakeholder community has little appetite to change the law.

International Jurisdictions

International patent offices are reviewing whether an AI agent should be eligible for inventorship, with the DABUS application as the first example. Several of the most popular patent offices have already concluded their legal frameworks will not recognize AI inventors. Specifically, the DABUS application was filed in multiple international patent offices, including the United KingdomEuropeIsrael, and as a Patent Cooperation Treaty (PCT) international application at the World Intellectual Property Organization (WIPO). Thus far, the European Patent Office (EPO) and the UK Intellectual Property Office (UKIPO) have refused the DABUS application on the grounds that an AI agent cannot be an inventor. It is likely that other patent offices will follow the EPO, UKIPO, and USPTO. Moreover, many jurisdictions appear disinterested in changing their law to recognize AI agents as inventors.

Inventor Considerations

For the foreseeable future, all patent applications should list only human inventors. Even if an inventor believes an invention was fundamentally conceived by an AI agent, there is likely at least one aspect of the invention that can be attributed to a human inventor. 

The threshold question is whether the human truly did nothing to conceive of the solution. Although an AI agent arrived at the particular solution, a human likely contributed to the conception in some fashion. Some examples of human involvement might include: 

  • inputting certain operational constraints or configurations to instruct the AI agent on its specialized operations; 

  • posing the problem to the AI agent through human operation or configuration; 

  • selecting and inputting specialized training data; and

  • creating some downstream operation or function that benefits from the AI-generated solution. 

The current state of AI technology has not yet reached general intelligence. It is thus highly likely that a human was involved in the conception or in the building of underlying technological features empowering an AI agent. Consequently, a human is also likely an inventor of the AI agent’s novel training or deployment algorithms or any novel downstream system operations enabled by the AI agent’s outputs. In filing patent applications related to the technology, the human that participated in the design of such an AI agent should thus be listed as the inventor. 

© 2020 Foley & Lardner LLPNational Law Review, Volume X, Number 121

TRENDING LEGAL ANALYSIS


About this Author

Austin Kim Intellectual Property Attorney Foley & Lardner
Associate

Austin J. Kim is an associate and intellectual property lawyer with Foley & Lardner LLP. He is a member of the firm’s Electronics Practice as well as a member of the Automotive and Technology Industry Teams.

Austin counsels clients in strategic patent portfolio development through various stages in procurement and across multiple jurisdictions. He has prepared and prosecuted applications in the electric vehicle space including fuel cells, autonomous driving, and remote controls. He also has experience in telecommunications, virtualization, cloud computing, web applications, and...

617-502-3295
Senior Counsel

Matthew Horton is a senior counsel and intellectual property lawyer with Foley & Lardner LLP. He is a member of the firm’s Electronics Practice. Matthew’s practice focuses on patent law and intellectual property protections, particularly in cybersecurity platforms and processes, artificial intelligence and machine learning, mobile devices and applications, cloud computing systems, and computer hardware.

With hands-on knowledge on software code and complex mathematics, Matthew is able to engage engineers, computer scientists and cybersecurity professionals at a technical level, and brings a combination of legal and technical knowledge easily tailored to address legal issues associated with high technology innovation. Matthew regularly counsels clients on developing strategic intellectual property portfolios, helping them to identify and develop business-oriented and industry-oriented portfolio strategies.

Matthew regularly helps clients with strategic IP portfolio development and patent litigation. In patent prosecution, Matthew has successfully drafted or prosecuted a variety of high-technology patents, including patent applications directed to cybersecurity technologies, artificial intelligence and machine learning, high-performance computing, high-availability infrastructures, cloud computing, mobile and web applications, mobile devices, business processes and database management, computer hardware, machine learning and semiconductors.

In addition to portfolio counseling, Matthew also represents large consumer electronics companies and financial institutions in patent litigation cases, in both traditional courts and AIA post-grant trial proceedings.

Matthew has extensive information technology and cybersecurity experience, serving as a systems engineer and cybersecurity engineer at Lockheed Martin and Electronic Warfare Associates. He played key roles on various cybersecurity and network engineering teams overseeing Federal network infrastructures. As a member of both cybersecurity and network engineering teams servicing agencies such as the Department of Homeland Security, the Department of Defense and the State Department, he contributed to anti-malware and anti-intrusion efforts, conducted anti-intrusion and network vulnerability assessments, and consulted on improving the cybersecurity postures for networks he supported.

Area of Focus

  • Strategic IP Planning
  • Patent Counseling and Prosecution
  • Patent Litigation
  • Inter Partes Reviews and Other Post-Grant Proceedings
  • Data and Software Rights, Patent Rights and Cybersecurity
  • Intellectual Property and Technology
202.295.4016