The U.S. Patent Office provides Inventorship Guidance for AI-Assisted Inventions

By Ryan N. Phelan on March 6, 2024

PatentNext Takeaway: The U.S. Patent and Trademark Office (USPTO) recently issued examination guidance regarding patentability for artificial intelligence (AI)-assisted inventions. The guidance states that AI-assisted inventions are not “category unpatentable.” Instead, when a natural person provides a “significant contribution” to an invention, such an invention can be patentable even if an AI system contributed to the invention. While the guidance does not constitute law, it is grounded in law, i.e., the Federal Circuit’s so-called Pannu factors, which serve as a test for ensuring that a natural person contributed, at least in part, to the conception of the invention as required in the Federal Circuit’s Thaler decision on AI inventorship. The guidance also provides several useful guidelines and examples to help patent practitioners determine what constitutes a “significant contribution” for purposes of establishing natural person inventorship and, thus, patentability for AI-assisted inventions.

****

The U.S. Patent and Trademark Office (USPTO) recently issued examination guidance titled “Inventorship Guidance for AI-Assisted Inventions” (referred to herein as the “USPTO Guidance”). The USPTO Guidance is directed to U.S. patent examiners for examining U.S. patent applications filed with the office. While the USPTO Guidance does not constitute law, it can be expected to impact how the USPTO can scrutinize inventions created with (or with the assistance of) artificial intelligence (AI).

Background

The USPTO Guidance was issued in response to President Biden’s Executive Order that called for guidance when AI was used by a natural person during the invention process. See PatentNext: Intellectual Property (IP) impacts from President Biden’s Executive Order on Artificial Intelligence (AI).

The USPTO Guidance was also expected in view of the prevailing “two schools of thought” regarding AI-assisted inventorship, which hinges on whether AI is perceived as merely a tool of a human inventor or as an inventor (or co-inventor) capable of human “conception.” See PatentNext: Do you have to list an Artificial Intelligence (AI) system as an inventor or joint inventor on a Patent Application?

The USPTO Guidance has adopted the latter of these “two schools of thought” by allowing an AI-assisted invention to be patentable so long as a natural person (a human) provided a “significant contribution” to the claimed invention.

Overview of the USPTO Guidance

The below sections summarize the major themes and important takeaways from the USPTO Guidance.

1. AI-Assisted Inventions Are Not “Categorically Unpatentable”

As a key takeaway, the USPTO Guidance provides that AI-assisted inventions are not “categorically unpatentable.” This means that AI-assisted inventions can be patented. However, at least one or more human inventors (“natural person(s)”) need to have “significantly contributed” to a given invention. The USPTO Guidance sums this up as follows:

[AI-assisted] inventions are not categorically unpatentable due to improper inventorship if one or more natural persons significantly contributed to the invention.

The USPTO Guidance requires at least one natural person to be listed as an inventor because it follows in the wake of the Federal Circuit’s decision in Thaler v. Vidal, holding “that only a natural person can be an inventor, so AI cannot be.” See PatentNext: Can an Artificial Intelligence (AI) be an Inventor?

2. The Human Contribution To The Invention Must Be “Significant”

The use of an AI system by a natural person does not preclude that natural person from qualifying as an inventor (or joint inventor) if the natural person “significantly contributed” to the claimed invention.

The USPTO Guidance reviews whether a human “significantly contributed” pursuant to the three Pannu factors (so-named after the Federal Circuit’s case in Pannu v. Iolab Corp., 155 F.3d 1344, 1351 (Fed. Cir. 1998)).

The Pannu factors are as follows:

Each inventor must contribute in some significant manner to the conception or reduction to practice of the invention.
Each inventor must make a contribution to the claimed invention that is not insignificant in quality, when that contribution is measured against the dimension of the full invention.
Each inventor must do more than merely explain to the real inventors well-known concepts and/or the current state of the art.

The Pannu factors are grounded in joint-inventorship law and are designed to probe whether a person significantly contributed to the conception of an invention so as to qualify as an inventor on a patent. The USPTO Guidance applies this law to AI-assisted inventions to probe whether a human significantly contributed to an AI-assisted invention for purposes of establishing at least partial human inventorship in view of the Thaler decision.

Thus, in the context of AI-assisted inventions, natural person(s) who create an invention using an AI system, or any other advanced system, must contribute significantly to the invention, as determined by the the Pannu factors.

3. Guiding Principles Regarding “Significant Contributions”

The USPTO Guidance provides several valuable comments and examples to determine whether a human made a “significant contribution” to a given invention. These include the following:

Input into an AI system, e.g., Prompt Engineering. A patent applicant can demonstrate a “significant contribution” by how a person constructs a prompt as entered into an AI system. The prompt can be designed in view of a specific problem to elicit a particular solution from the AI system (Guiding Principle 2). The USPTO Guidance, therefore, spotlights that good practice is to keep a log of prompt entries into AI systems should the need arise to prove inventorship based on prompt engineering. See PatentNext: Artificial Intelligence (AI) Policy Considerations. On the other hand, a natural person who only presents a problem to an AI system may not be a proper inventor or joint inventor of an invention identified from the output of the AI system.
Output of an AI system, e.g., adding to AI output, can lead to a “significant contribution.” A person who takes the output of an AI system and makes a significant contribution to the output to create an invention may be a proper inventor (Guiding Principle 3). Alternatively, in certain situations, a person who conducts a successful experiment using the AI system’s output could demonstrate that the person provided a significant contribution to the invention even if that person is unable to establish conception until the invention has been reduced to practice. However, a natural person who merely recognizes and appreciates the output of an AI system as an invention, particularly when the properties and utility of the output are apparent to those of ordinary skill, is not necessarily an inventor.
Construction of a specific AI system to solve a specific problem. In some situations, a natural person who designs, builds, or trains an AI system in view of a specific problem to elicit a particular solution can qualify as an inventor, where the designing, building, or training of the AI system is a significant contribution to the invention created with the AI system (Guiding Principle 4). However, maintaining “intellectual domination” over an AI system does not, on its own, make a person an inventor of any inventions created through the use of the AI system (Guiding Principle 5). Therefore, a person simply owning or overseeing an AI system that is used in the creation of an invention without providing a significant contribution to the conception of the invention does not make that person an inventor.

4. Example Claims as provided by the USPTO Guidance

The USPTO Guidance also provides two examples to assist the public and examiners in the application of the above-mentioned Guiding Principles in specific situations. The examples are hypothetical and show the application of the Guiding Principles to several scenarios, some of which demonstrate inventorship and some of which fall short of inventorship, each ultimately in view of the Pannu factors.

First example (Electromechanical example)

The first example is in the electromechanical space and is titled “Transaxle for Remote Control Car.” It involves a hypothetical scenario where two engineers (Ruth and Morgan) at the “XYZ Toy Company” use AI to develop remote-controlled toy cars. They use an AI system named “Puerto5” that receives natural language prompts as input and generates text, images, and other media as output (e.g., not unlike OpenAI’s GPT-4).

In some scenarios, Ruth and Morgan are not considered inventors. These include when they claim well-known features of transaxles (e.g., use of steel or aluminum) or provide input to prompts into Puerto5 that are general in nature and ask Puerto5 to solve the problem, e.g., “Create an original design for a transaxle for a model car, including a schematic and description of the transaxle.” In this latter scenario, the patent claims developed based on Puerto5’s output are not patent-eligible because, according to the example, they failed to make any significant contribution to the conception of the invention (the problem) to be solved given the way the prompt was constructed. Instead, Ruth and Morgan rely on the output of Puerto5 alone, and, thus, do not provide a “significant contribution.”

In another scenario (Scenario 3), they use the output of Puerto5 as a starting point to develop a new design, which includes designs that they themselves conceived (e.g., elongating the casing of the transaxle and using removably attachable transmission fasteners). Such additional contributions, which stem from the conception of natural persons, exemplify “significant contributions” that, when added to patent claims, make the patent claims eligible despite the assistance from the AI system Puerto5.

Second example (Pharmaceutical example)

The second example is in the pharmaceutical space and is titled “Developing a Therapeutic Compound for Treating Cancer.” The example involves a hypothetical where Marisa, a professor, develops a drug to treat prostate cancer. She wants to use AI to identify lead drug compounds for prostate cancer therapy that selectively target the mutated androgen receptor protein (AR). In the example, she uses an AI system named DTIP that accepts drug-target pairs (in silico data) as inputs and that outputs numerical value(s) representative of the binding affinity for a given pair. She identifies six (6) drug compounds CID_1 to CID_6 based on providing input to DTIP from publicly known datasets (approximately 20,000 compounds). CID_1 has the highest predicted binding affinity between the compounds and a mutated AR receptor protein.

Marisa and Naz, her postdoctoral fellow, then conduct wet lab experiments on CID_1 to determine that it binds to both mutated and unmutated AR. They create a modified version (CID_1-mod) that has a more selective binding to the mutated AR and shows anti-tumor potency. The CID_1-mod becomes the subject matter of a patent claim. Maria and Naz are considered inventors, and the patent claim is patent-eligible because of the significant contributions that Maria and Naz conceived during the web lab experiments.

However, other actors, including Lauren, who designed the DTIP AI system, and Raghu, who assisted Maria in using DTIP to identify the original CID_1 to CID_6 drug compounds, are not inventors. Neither provided a significant contribution; Lauren simply designed a general AI system, not a specific AI system for solving a specific problem. Raghu simply assisted with inputting data into DTIP, which the USTPO GUdiance considers an insignificant contribution to the overall invention.

Finally, in a second scenario (Scenario 2), Raghu assists Marisa in designing a new AI system named Molecule Optimizer (MO). Importantly, Raghu works with Marisa to design MO as a specific AI system designed to solve a specific problem, i.e., identifying absorption, distribution, metabolism, excretion, and toxicity (ADMET) properties in the CID_1 to CID_6 drug compounds and developing modified versions that improve these properties. Once trained in the specific manner provided by Raghu and Marisa, MO is able to output modified versions of CID_1 to CID_6, so-called MID_1 to MID_6, which become the subject matter of a patent claim. Because MO was a specific system trained and designed by Raghu and Marisa to solve a specific problem (e.g., identifying modified drug compounds for ADMET purposes), Raghu and Marisa are considered inventors because they contributed significantly to the specific AI system.

5. Each Patent Claim Must Include A “Significant Contribution” By A Natural Person

The USPTO Guidance requires that a natural person must have provided a “significant contribution” (per the Pannu factors) to every single claim of a patent invention. Said another way, a claim wholly created by an AI tool is not patentable because that would violate the Thaler decision, which prohibits the naming of AI systems as inventors.

The USPTO Guidance states that such determination is highly fact-specific and is to be made on a claim-by-claim basis. Examiners are instructed to reject claims, under 35 U.S.C. 101 and 115, for any claim that is determined “from the file record or extrinsic evidence that at least one natural person, i.e., one or more named inventors, did not significantly contribute” to a given a claim.

Presumably, dependent claims that are wholly “conceived” by an AI tool, but that nonetheless depend on a claim that includes at least one “significant contribution” of a natural person, would be patentable.

6. An AI System Cannot be Listed As An Inventor, Even If The AI System Was “Instrumental” In The Creation Of An Invention

An AI system may not be listed as an inventor. The USPTO Guidance makes clear that applications and patents must not list any entity other than a natural person as an inventor or joint inventor, even if an AI system may have been instrumental in creating the claimed invention. According to the USPTO Guidance, “[t] his position is supported by the statutes, court decisions, and numerous policy considerations.”

7. No AI Reporting Requirement

The USPTO Guidance states that it is not changing or modifying its requirements regarding a patent applicant’s duty of disclosure or duty of inquiry. Thus, a patent applicant does not have an affirmative duty to disclose whether an invention was created with the assistance of an AI tool.

This is the opposite of the Copyright Office’s requirement that a copyright registrant must disclose AI-generated works or portions thereof to the Copyright Office when seeking to register the work if it was generated, at least in part, with an AI system or tool. See PatentNext: U.S. Copyright Office Partially Allows Registration of Work having AI-generated Images (“Zarya of the Dawn”).

The USPTO Guidance, however, cautions that, if in doubt with respect to AI-assisted inventions, a patent applicant should confirm that the claims of a given patent include a significant contribution of a natural person by determining that the claims satisfy the Panuu factors, e.g., “Given the increasing use of AI systems in the invention-creation process, applicants should take extra care in ensuring each named inventor in a patent application or patent provided a significant contribution to a claimed invention as described by the Pannu factors.”