Can artificial intelligence-assisted inventions be eligible for patent protection? This is the central question many courts, agencies and human inventors are trying to resolve.
At the intersection of technological advancement and intellectual property law, AI-assisted inventions, including generative AI, challenge the traditional concept of who gets credit for an invention and to what extent they can protect their innovations.
To resolve this, the intent and history of patent protection needs to be understood and how those precedents might apply in light of developing technology should be considered.
The U.S. patent system is rooted in Article 1, Section 8, Clause 8 of the Constitution, which empowers Congress "to promote the progress of science and useful arts, by securing for limited times to authors and inventors the exclusive right to their respective writings and discoveries." The governing law for patents is codified in Title 35 of the U.S. Code and is overseen by the federal agency, the U.S. Patent and Trademark Office. The USPTO is responsible for granting patents as well as examining, issuing and developing regulations relating to the patent process.
The USPTO grants three types of patents: utility patents for new and useful processes, machines, manufactures, or compositions of matter; design patents for new, original and ornamental designs for manufactured articles; and plant patents for new varieties of asexually reproducing plants. Utility and plant patents generally have a term of 20 years from the filing date, while design patents last 15 years from the grant date.
Patents are more difficult to obtain than copyright or trademarks because the system is built to reward genuinely novel and nonobvious inventions, setting a high bar for innovation rather than rewarding minor changes to existing technology.
There are a few basic principles regarding patent eligibility and inventorship in the U.S.:
- Only "natural persons" can be inventors on U.S. patents.
- For utility patents, the invention must fall within one of the four statutory categories — process, machine, manufacture or composition of matter.
- The invention must be novel, nonobvious and useful to qualify for patent protection.
- The patent application must describe the invention sufficiently to enable others to make and use it.
In the U.S., various cases and standards have impacted the practices and requirements around patents. Of particular note for purposes of this discussion, the "significant contribution" standard from Pannu v. Iolab quantifies the amount of involvement by a particular person required during the inventive process to qualify for credit as an owner or co-owner. Levels range from low involvement, where there is no significant contribution, to medium involvement, where there is likely some form of significant contribution, to high involvement, where there is a significant contribution.
The contribution must be to the conception of the invention, which is the formation of a definite and permanent idea of the complete and operative invention. Mere assistance or peripheral involvement in the invention process does not meet the threshold for co-ownership or ownership of a patent.
How does AI change patents?
In October 2023, U.S. President Joe Biden issued an executive order on the "safe, secure, and trustworthy development and use" of AI, which, requires the USPTO to publish guidance on AI's inventorship and patentability issues, among other things. The directive aims to guide the application of the Pannu inventorship standard to modern situations where technology like AI might play a role in the inventive process.
In February 2024, the USPTO issued such guidance to inform the public on the level of human involvement needed to satisfy the 26-year-old "significant contribution" standard. It laid out five guiding principles to determine whether a natural person's contribution is considered sufficiently significant when it comes to an AI-assisted invention:
- "A natural person's use of an AI system in creating an AI-assisted invention does not negate the person's contributions as an inventor. The natural person can be listed as the inventor or joint inventor if the natural person contributes significantly to the AI-assisted invention."
- "Merely recognizing a problem or having a general goal or research plan to pursue does not rise to the level of conception." A natural person who merely presents a problem to an AI system may not qualify as an inventor, but may demonstrate a significant contribution by carefully crafting the prompt to direct the AI system to generate a specific solution.
- "Reducing an invention to practice alone is not a significant contribution that rises to the level of inventorship." A natural person who merely recognizes an AI system's output as an invention is not necessarily an inventor. However, if the person contributes significantly to the output or conducts an experiment that contributes to the invention, they may qualify as a proper inventor.
- "A natural person who develops an essential building block from which the claimed invention is derived may be considered to have provided a significant contribution to the conception of the claimed invention even though the person was not present for or a participant in each activity that led to the conception of the claimed invention." In some cases, a natural person who designs, builds or trains an AI system to solve a specific problem could be considered an inventor if their contribution is significant to the resulting invention.
- "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 an AI system." Owning or overseeing an AI system used in creating an invention does not make a person an inventor unless they significantly contribute to the invention's conception.
These guiding principles critically assess the extent and significance of a natural person's contribution to the development and process of an AI-assisted invention. The USPTO seeks to clarify the issues of inventorship in the context of AI to ensure inventorship reflects meaningful human input rather than mere oversight of AI processes.
Patent activity in pharma development
The pharmaceutical industry is one example of how AI is being used to assist human inventors, particularly in areas such as drug discovery, generating new formulas and predicting a drug's potential side effects or effectiveness. Researchers and pharmaceutical companies are developing AI-driven platforms to identify disease-related targets, predict drug interactions and design experiments.
With this AI-assisted analysis, drug discovery is becoming faster and more focused because researchers can target specific disease mechanisms instead of testing drugs randomly. Machine learning models can predict how a drug will behave in the body and its potential harmful effects, which is essential to help determine if a drug is safe and effective without extensive animal testing. Using AI analysis, scientists can develop personalized medicine using real-world patient data, potentially even to tailor treatments to individual patients.
One of the most notable companies using AI for drug discovery is Atomwise, which uses AtomNet, a deep-learning technology that predicts the binding of small molecules to proteins. Binding is the process of the drug interacting with a specific protein related to a disease. Typically, drug discovery would take physically testing thousands of molecules in a lab, which is time-consuming and costly. AI helps researchers identify the specific compounds most likely to target the disease faster and be the most promising drug candidates.
How other nations are navigating AI inventorship and patentability
Japan, China and the EU, among others, have begun issuing guidance on AI-related IP issues, each taking different approaches to ensure clarity and fairness in how AI inventions may be patented.
In Japan, patents are governed by the Patent Act of 1959. Regarding AI patent eligibility, Japan has updated its patent application examination requirements, showcased by the definition of "AI-related invention" and has seen a notable increase in domestic applications, particularly in AI-core technology.
The Japan Patent Office has updated case examples to clarify examination procedures, covering various technical fields and functions tied to AI. The description requirements for AI-related inventions "include inventions taking advantage of AI-related technology in various technical fields and inventions of a product which is presumed to have a certain function because of AI." However, this description alone will not satisfy the requirements without evaluating the product's function, which has been made "unless an estimation result by AI can be a substitution for an evaluation on a product that has actually been made."
By refining the examination standards and incorporating AI-specific criteria, Japan seems to be encouraging domestic innovation and the use of AI technologies while ensuring its IP framework can keep pace with the complexities of AI.
In China, patent guidelines have been provided to clarify the eligibility criteria for AI-related inventions. Chinese patent law has established that inventions must involve "technical means, solve technical problems, and achieve technical effects" to be patentable. AI algorithms or mathematical rules are typically excluded from patent protection under Article 25 of the Patent Law, as they are considered "rules and methods for mental activities."
However, an invention relating to computer programs can be patentable if it involves executing programs to solve technical problems, uses technical means through computers executing programs to control and process external or internal objects, or obtains technical effects.
The recent updates reflect China's intent to treat AI-related inventions like other computer-program inventions, as long as they meet the technical requirements for patent eligibility.
Meanwhile, the EU also considers AI inventions part of "computer-implemented inventions" and part of the list of exclusions under mathematical methods. Computer programs, models and algorithms are typically considered part of mathematical procedures or abstract activities that lack a technical character. Therefore, AI inventions must demonstrate a "technical character" beyond just being a computer program to be patentable.
The European Patent Office may grant a patent in use cases when AI moves beyond the abstract and is applied to address a specific technical problem. For example, using a neural network in a heart-monitoring device to detect irregular heartbeats demonstrates a technical contribution. Even with this level of specificity, however, European inventorship requirements, like those in the U.S., require the named inventor to be a natural person with legal capacity.
Conclusion
As AI inventions advance and reshape industries worldwide, they also create important new challenges for patent law and IP protections.
In the U.S., the USPTO is following its core patent standards of the significant human contribution element, ensuring there is still a fundamental level of human creativity involved.
Japan, China and Europe are taking similar steps by updating their patent frameworks to include AI-related inventions and considering the technical contributions and the role of natural persons in the evaluation process.
These areas and others will still need to balance technological advancements with current legal standards, ultimately determining who ― or what ― can receive patent protection.
Natalie Linero is a research associate at Luminos Law.
