Patenting RNA Therapeutics In The Age Of AI: What Companies Must Get Right Early

By Robert S. Jacobson and Niko Moses, Ph.D., Marshall, Gerstein & Borun LLP

Artificial intelligence (AI) is rapidly transforming the discovery and engineering of RNA-based therapeutics. Modern AI platforms can design optimized mRNA sequences, generate potent siRNAs, and identify novel RNA targets. Yet patent law has not kept pace with these technological advances. Although the litigation landscape is rich with disputes over mRNA vaccines, RNAi, antisense oligonucleotides, and delivery systems, courts have not yet addressed a fundamental question: how will patent law treat RNA therapeutic inventions conceived with the assistance of increasingly autonomous AI systems?

The absence of clear legal precedent places companies in a precarious position. Many of the foundational doctrines governing patent validity were developed for inventions created by human inventors using conventional research tools, not for a world in which generative AI can propose thousands of RNA sequences in seconds or machine learning systems can navigate chemical and biological designs too numerous for human intuition alone.

As a result, innovators risk building valuable therapeutic portfolios on patents that could face future challenges if AI’s role in the inventive process is not properly considered. Questions related to inventorship and obviousness, in particular, may become increasingly important as AI becomes more deeply integrated into the therapeutic development pipeline.

Establishing Human Inventorship In AI-Assisted RNA Innovation

Inventorship is the first doctrinal hurdle for RNA therapeutics generated using AI models. Under U.S. patent law, inventorship turns on conception, defined as the formation in the mind of the inventor of a “definitive and permanent idea of the complete and operative invention.” Because only natural persons can be inventors, AI systems, regardless of their sophistication, cannot be named on a patent.

This creates a fundamental tension: AI systems can now generate therapeutic RNA sequences with minimal human input. If a researcher merely accepts an AI-generated output, that human may not have conceived of the invention as defined by U.S. law. In this case, the AI system may need to be listed as an inventor to patent the AI-generated sequence, which U.S. patent law currently prohibits, and the company may be unable to protect the sequence via patents.

Humans can still qualify as inventors when they contribute the technical insights that shape or define the invention. Because inventorship is assessed on a claim-by-claim basis, patent applicants should carefully draft claims to reflect meaningful human contributions to the inventive process.

The key question becomes identifying those contributions. In the context of AI-assisted RNA therapeutics, they may come from computer scientists who design and train the AI systems, biologists who define the underlying therapeutic objectives and constraints, or in some cases, both.

How Computer Scientists Establish Inventorship

A computer scientist may qualify as an inventor when their contribution goes beyond merely operating an AI tool and instead helps shape the conception of the claimed RNA therapeutic. In the context of AI-assisted RNA development, inventorship often turns on whether the individual made meaningful decisions that directed the AI system toward the ultimate invention.

Examples of potentially inventive contributions include:

1. Designing or modifying the AI model. If the computer scientist adjusts the model architecture, loss functions, training data, or constraints specifically to enable the AI model to produce the claimed RNA sequence or structure, this may constitute conception. Examples include engineering a model to optimize untranslated regions (UTRs) for stability in a particular cell type, implementing constraints that produce a desired RNA sequence, or designing a generative model capable of producing a certain RNA structure, such as a circular RNA. In these situations, the inventive contribution is not the AI model itself but the human-directed decisions that caused the model to generate the invention.
2. Defining the rules that guide sequence generation. Inventorship may also arise when a computer scientist establishes the parameters that govern how the AI system evaluates and generates candidate RNA sequences. For example, a researcher may specify thermodynamic thresholds, off-target scoring rules, immunogenicity filters, structural motifs, or ranking metrics for evaluating potential candidates. These rules guide the AI model toward a particular solution space and can reflect the human inventor’s conception of the invention.
3. Integrating AI-generated components into a therapeutic design. Even when an AI system generates part of an RNA sequence, a computer scientist may still qualify as an inventor by contributing to the overall therapeutic design. Examples include combining an AI-generated mRNA with a specific delivery system optimized through computational modeling for targeted delivery to a specific tissue, designing a multicomponent RNA circuit using algorithmic simulations to ensure proper signal propagation, or engineering a self-amplifying RNA backbone around an AI-generated coding region by determining the optimal arrangement of elements to maximize expression. In these situations, the inventive contribution may lie in the way the AI-generated component is incorporated into a broader therapeutic strategy.

How Biologists Establish Inventorship

Biologists often play a central role in AI-assisted RNA development by defining the biological objectives, constraints, and scientific judgments that guide the inventive process. Although an AI system may generate candidate sequences, a biologist may qualify as an inventor when their expertise shapes the conception of the claimed therapeutic.

Defining the Therapeutic Objective

A biologist may contribute to conception by identifying the biological problem the invention is intended to solve. This may include selecting the gene to silence, the exon to skip, the protein to express, or the cell type to target. These decisions establish the scientific framework within which the AI system generates and evaluates candidate RNA sequences. In many cases, they serve as the foundation for the eventual invention.

Establishing the Biological Constraints

Inventorship may also arise when a biologist defines the biological requirements that an AI-generated sequence must satisfy. For example, a biologist may require that a candidate sequence avoid activating an innate immune response, achieve nuclear localization to modulate splicing, maintain a particular secondary structure to improve translation efficiency, or meet acceptable off-target recognition profiles. These constraints materially influence the AI-generated outputs in a biologically relevant manner.

Applying Scientific Judgment to AI-Generated Outputs

A biologist may also qualify as an inventor when they use their expertise to evaluate, refine, or select among AI-generated candidates. Examples include selecting a final therapeutic sequence from among AI-generated candidates based on biological performance, modifying a sequence to improve stability, potency, or specificity, or identifying unexpected biological properties associated with the output sequence.

When An Individual Is Not An Inventor

Just as failing to identify a true inventor can jeopardize a patent, incorrectly naming a non-inventor can render a patent vulnerable. Thus, it is equally important to identify activities that do not confer inventorship, which include running an AI model without contributing to conception, maintaining computational infrastructure, performing routine wet-lab validation, supervising a team without contributing technical insight, funding the research, or managing the research.

Practical Guidelines For Companies

Patent applicants face increased scrutiny regarding the use of AI in their development process. To prepare, companies should develop policies to document the role of both AI systems and human contributions throughout the innovation process. Clear records can help support inventorship determinations, strengthen patent applications, and reduce the risk of future challenges.

In particular, companies may consider policies that maintain contemporaneous records identifying who contributed specific ideas and concepts, document human decision-making at each stage of AI-assisted design, record the rationales for selecting or refining AI outputs, track the respective contributions of all involved in the development process, and preserve evidence demonstrating how human expertise shaped the conception of the claimed invention.

Companies should also recognize that inventorship analysis may become more complex as AI systems assume a larger role in therapeutic discovery. Ultimately, the strongest patent positions will likely be held by organizations that treat AI not as a substitute for human inventorship but as a powerful tool operating within a carefully documented human-led innovation process.

Rethinking Obviousness Standards

Under U.S. law, an invention is unpatentable if, at the time of filing, it would have been obvious to the skilled artisan in view of the prior art. Historically, RNA therapeutics benefited from the field’s inherent unpredictability: small changes in sequence or structure often produced large, unexpected differences in biological activity. However, AI-accelerated discovery threatens to erode that unpredictability by making certain design choices appear more systematic and more predictable, and therefore more likely to be considered obvious.

One central question facing companies is whether the use of AI to generate and rank thousands of candidate RNA sequences will make resulting inventions appear more predictable. AI can identify optimal siRNA seed regions, rank mRNA constructs by predicted translation efficiency, and propose sequences with minimal off-target effects. As these capabilities become more sophisticated, challengers may argue that certain AI-generated sequences represent the expected outcome of applying known computational techniques rather than a nonobvious advance in the field.

AI-generated sequences may be vulnerable to obviousness challenges when users provide the AI system with well-known design rules that are not adapted to a particular inquiry. To improve positioning with respect to potential obviousness challenges, applicants can affirmatively demonstrate that their AI-generated RNA therapeutics exhibit unexpected properties or required non-routine scientific judgment to arrive at the AI output.

These strategies include:

1. Highlighting unexpected results. Provide and explain experimental tests that demonstrate the unexpected characteristics of the AI-generated RNA therapeutic. Examples may include unusually high potency, improved stability or half-life, unexpected tissue specificity, reduced off-target immunogenicity, or other biological properties that would not have reasonably been predicted from the prior art.
2. Explaining why the AI output was not predictable. Example explanations should articulate why the sequence was not an obvious variation of a prior art sequence, how AI was used to explore a vast or unconventional design space, how the AI model was adapted to provide the specific output, and/or why a skilled artisan would not reasonably have expected success. Such a narrative can help distinguish the invention from what are otherwise obvious variations of the prior art.
3. Claiming features that reflect inventive insight. Carefully consider a claim strategy to focus on human contributions, such as identifying specific structural motifs, unique secondary-structure elements, unexpected chemical modification patterns, and/or novel combinations of sequences and delivery features.
4. Avoiding overly broad claims. Applicants should be cautious about claims that encompass all sequences generated by the model (or all sequences meeting a certain computational score), which invite challenges of obviousness.

Conclusion

Acceleration of RNA-based therapeutic design using AI tools offers enormous promise for companies, but incorporation of these AI tools into the research pipeline may require changes to the product development process to ensure their IP rights are not harmed. By proactively addressing inventorship and obviousness challenges, companies can secure durable patent protection and confidently protect advances in the next generation of RNA-based medicines.

About The Authors:

Robert S. Jacobson is a partner and patent lawyer at Marshall, Gerstein & Borun LLP, Chicago’s largest intellectual property law firm. He counsels innovative companies on a broad range of intellectual property matters, combining his legal experience with five years of prior engineering work in the telecommunications industry. He advises clients on patent drafting and prosecution, product strategy, infringement analysis, open-source software compliance, and portfolio management, with a practice focused on technologies such as digital communications, wireless devices, and telecommunications. Before becoming an attorney, Rob worked as a software engineer at Motorola and later as a device engineer at U.S. Cellular, where he helped ensure mobile devices met technical requirements for wireless networks. That hands-on engineering experience enables him to quickly understand complex technologies and provide practical, business-focused advice tailored to his clients' objectives.

Beyond his practice, Rob is active in Chicago’s IP community through the Chicago Intellectual Property Alliance’s mentoring program. He was recognized in the 2026 edition of Best Lawyers: Ones to Watch in America for Patent Law.

Niko Moses, Ph.D., is a patent attorney at Marshall, Gerstein & Borun LLP. She helps biomedical innovators protect their inventions through strategic patent prosecution, with experience in patent drafting, prior art research, and developing effective responses to patent office actions. Her practice focuses on biotechnology and life sciences, where she combines scientific knowledge with practical legal counsel to help clients build and strengthen their intellectual property portfolios. Before entering the legal profession, Niko earned a Ph.D. in cancer biology and conducted oncology research, giving her firsthand insight into the scientific and technical challenges faced by life sciences companies. That background enables her to quickly understand complex biomedical innovations and communicate effectively with inventors and researchers. She was recognized in the 2026 edition of Best Lawyers: Ones to Watch in America for Patent Law.

DISCLAIMER: The information contained in this article is for informational purposes only and is not legal advice or a substitute for obtaining legal advice from an attorney. Views expressed are those of the author and are not to be attributed to Marshall, Gerstein & Borun LLP or any of its former, present, or future clients.