The emergence of artificial intelligence (AI) in drug development has significant ramifications for the pharmaceutical industry, patent law, and potential patent challengers. As companies increasingly rely on AI to identify promising drug candidates and streamline development processes, understanding how this technology intersects with patent law is crucial for anyone who is looking to challenge effectively the patents that will inevitably result.

Revolutionizing Drug Development

An article from The Economist last year highlighted how AI tools are transforming the landscape of drug discovery. The potential to reduce the time and costs of preclinical drug development by 25-50% marks a significant leap forward. Noteworthy examples include BenevolentAI’s use of AI to identify baricitinib as a potential COVID-19 treatment. Companies like Genentech and Insilico Medicine have publicized their efforts to employ AI in the drug identification and development space. Insilico in particular has been very vocal about its success developing “the first AI-discovered and AI-designed drug.” And these are just the efforts that companies are choosing to publicize; it is a safe bet that it's just the tip of the iceberg.

These advancements raise interesting and often novel questions about the patentability of AI-assisted inventions and the legal complications they pose, particularly for those who might seek to challenge such patents in the future.

35 U.S. Code §§  115 and 116 Inventorship Issues

What is a “significant contribution”?

In response to the advancement of AI, the United States Patent and Trademark Office  issued guidance regarding AI and inventorship. This guidance reiterates that inventors must be natural persons but allows for the contributions of AI in the inventive process.

The guidance emphasizes that each human inventor must have made a “significant contribution” to the invention's conception—not merely acknowledged the AI's output. A significant contribution could involve refining the AI's outputs or validating its hypotheses through experimental work. This raises crucial questions about inventorship—if AI identifies a novel compound and its potential use to treat a particular disease, what contribution is left for a human to make, at least with regard to the patent on the initial compound and any method of using it? The PTO came close to evaluating this scenario in one offered example, but found inventorship was proper in the hypothetical situation where the purported inventors further modified the computer-output compound. The much closer question will be where the AI identifies the specific compound that is ultimately patented.

It is worth noting that, while this guidance remains in effect, the Executive Order that prompted it has been rescinded, so expect this to be an evolving space.

Applicants' Disclosure Obligations

The PTO's guidelines require applicants to disclose information that verifies each inventor's significant contributions to the invention, but the PTO has made clear that it does not view AI as having materially changed this requirement, writing:

The USPTO is not changing or modifying its duty of disclosure. However, applicants and patent owners are reminded of their existing duty of disclosure and its applicability to the inventorship determination. . . . The USPTO does not believe this inventorship guidance will have a major impact on applicants' disclosure requirements. However, special care should be taken by those individuals subject to this duty to ensure all material information is submitted to the USPTO to avoid any potential negative consequences.

Thus, it is likely that applicants will not volunteer any more information than is necessary regarding any role of AI in the invention.

For those seeking to challenge a patent, this adds a layer of complexity: litigants will likely need to undertake extensive discovery to gather evidence demonstrating that AI, and not the named inventors, should be deemed responsible for the invention. If challengers can present compelling evidence that the named human inventors did not provide a significant contribution, they may successfully invalidate the patent based on improper inventorship.

35 U.S.C. § 103 Obviousness Issues

Redefining the Person of Ordinary Skill in the Art

As AI rapidly reshapes drug discovery, the definition of the person of ordinary skill in the art is also evolving. As the PTO has recognized and asked for comment regarding, the baseline skill and knowledge of a POSA may need to evolve and expand to include the use of AI tools. Perhaps the definition of a person of ordinary skill in the art does not require a human person at all.

AI-Generated Prior Art

Another pressing question that arises is how to treat AI-generated outputs as prior art. Given AI's ability to synthesize vast quantities of biological and chemical data, its output may serve as a vast repository of prior art that challengers can leverage in patent disputes. Determining whether an invention is non-obvious may become more challenging as AI fills the landscape with novel compounds. Questions will inevitably arise as to whether AI-generated prior art sufficiently enables any challenged inventions.

AI-Derived compounds

Another consideration is whether it is obvious to input into an AI model a starting compound and ask the model to identify promising modifications that might address a particular disease, reduce potential side effects, or improve efficacy. But if the AI generates novel and unique responses each time (as many of the most popular models do), how can a challenger ever prove what output any particular model would have produced as of the priority date? Perhaps it will be true that the inventor's particular (and likely proprietary) AI model generated that output, but “patentability shall not be negated by the manner in which the invention was made.” 35 U.S.C. § 103. Potential litigants and their counsel should give serious thought to these questions that will eventually present themselves.

Conclusion

The integration of AI into drug development not only reshapes the pharmaceutical landscape but also the legal frameworks governing it. The interplay between AI and patent law will surely continue to evolve, creating new challenges and opportunities for all those engaged in the pharmaceutical industry. For those considering challenging patents resulting from AI-generated innovations, a comprehensive understanding of these issues will be critical.