AI-designed drug patent dispute exposes limits of human-only inventorship
A discrepancy between Insilico Medicine’s public claims and its patent filing highlights the legal tension between rapid AI-driven discovery and traditional intellectual property rules.
Biotech company Insilico Medicine has drawn attention to a growing legal ambiguity in intellectual property law by publicly attributing the discovery of a new drug for pulmonary fibrosis to its generative AI platform. In a press release, the company stated that the molecule was “discovered by” its artificial intelligence system, a claim that underscores the increasing role of AI in generating drug designs that might otherwise elude human researchers.
However, when Insilico filed for a patent to protect the new chemical structure, the application named five human employees as the inventors, including CEO Alex Zhavoronkov. The filing made no mention of the AI platform, a move that aligns with current US statutes but creates a notable gap between public marketing and legal documentation.
This discrepancy stems from a 2022 ruling in the DABUS case, where US courts concluded that an inventor must be an “individual,” a term interpreted to mean a human being. Because machines are not people, they cannot currently be named as inventors under US law. Consequently, even if AI plays a fundamental role in a discovery, only humans can take legal credit for the invention.
Zhavoronkov defended the inclusion of human inventors by noting that human chemists are still required to synthesise the drugs, create variants, and test them on animals. He argued that even if the process were fully roboticised, a human would still need to “push the button and give the budget,” thereby maintaining a human contribution to the work.
The US Patent and Trademark Office (USPTO) has recently reversed its administrative guidance, now treating AI as a mere tool, similar to a calculator, rather than a co-inventor. Under this current stance, applicants are not required to explicitly mention AI in their filings, provided that a human is identified as the inventor.
Legal experts warn that this approach may face future challenges. Ryan Abbott, a partner at Brown, Neri, Smith & Khan, who brought the original DABUS case, noted that patents could be invalidated if they are found to list the wrong inventors. He suggested that if US policy excludes AI-generated outputs from direct protection, it could potentially hinder future drug development, raising questions about whether simply “pushing a button” constitutes sufficient inventorship.
