The pharmaceutical industry, which is actively integrating artificial intelligence into drug development, has encountered a fundamental legal barrier in the United States. Current legislation requires patent applications to list only human inventors, creating a paradoxical situation for companies where neural networks play a key role in creating molecules.
The example of Insilico Medicine is telling: the company announced that its generative AI platform had produced a potentially effective compound against pulmonary fibrosis. In the filed application, artificial intelligence is not mentioned at all—five people are listed as inventors, including CEO Alex Zhavoronkov. This is not a coincidence but a forced measure dictated by case law.
The key ruling came in 2022 in the DABUS case, when an appellate court in Washington determined that, by law, only a human can be an inventor. The term "individual" in U.S. regulations is unambiguously interpreted as a natural person. Since then, patent disputes have shifted toward proving the "sufficiency of human contribution"—the formal involvement of specialists in the synthesis, modification, and testing of molecules.
The situation is compounded by regulatory policy uncertainty. The U.S. Patent and Trademark Office (USPTO) initially issued guidelines on criteria for co-authorship when using AI, but after a change in administration, it shifted course, viewing neural networks as "a tool like a calculator" that does not require disclosure in an application. This approach, in my view, merely postpones the inevitable: as AI autonomy grows, it will become impossible to ignore its role in inventions.
Ryan Abbott, a partner at the law firm Brown, Neri, Smith & Khan, predicts a wave of patent challenges if errors in describing human involvement come to light in documentation. His position is telling: "If I asked Claude to cure cancer, it would be inappropriate to claim that it was my achievement."
Sarah Korman, chief legal officer at Isomorphic Labs, is confident that current law will need to be revised as AI's role strengthens. For now, companies are forced to artificially keep humans in the development chain, which creates risks for the entire industry. It is worth recalling that Anthropic recently announced the launch of the Claude Science environment for scientists, stating plans to develop drugs for "neglected" diseases—this will only intensify pressure on the outdated patent system.
Analytical conclusion: The current legal vacuum is not merely a bureaucratic obstacle but a systemic brake on investment in AI-driven pharmaceuticals. Until legislation recognizes AI as an inventive entity (or, at the very least, establishes clear criteria for co-authorship), we will witness either the concealment of real data or legal battles that drag on for years. The market needs predictability; otherwise, U.S. leadership in this field will be lost.