AI-Driven Drug Discovery: Who Gets the Patent Credit?

The press release was clear and triumphant: a new drug candidate for pulmonary fibrosis had been discovered by artificial intelligence. Yet the patent application told a different story. There, the inventors listed were five human beings. This quiet pivot by biotech firm Insilico Medicine isn’t a mistake or an oversight; it’s a legal necessity that cuts to the very heart of who—or what—gets credit for an idea in the 21st century.

Our current intellectual property system, a framework built for human ingenuity, is facing an unprecedented test from generative AI. These systems are no longer just tools for analysis; they are generative partners capable of proposing novel molecular structures that a human chemist might never conceive. As Sarah Korman, a patent attorney now with Alphabet’s Isomorphic Labs, noted at an MIT Technology Review event, the law is unequivocal for now: there must be a human inventor, or there is no invention and no patent. This stance was cemented in 2022 when a US appeals court ruled that an AI system named DABUS could not be listed as an inventor on a patent, as the statute defines an inventor as an “individual”—a human being.

The philosophical questions this raises are profound. If an AI platform autonomously designs a viable, novel drug molecule, who truly owns that flash of insight? The developers who built the model? The scientists who framed the initial problem? Or the machine itself? Legal scholar Ryan Abbott, who brought the landmark DABUS case, argues that our patent laws exist to “promote the Progress of Science and the Useful Arts,” a clause rooted in the US Constitution. If AI-generated outputs are excluded from protection, he warns, it could inadvertently stifle the very innovation the system is meant to encourage. The concern is that without the economic incentive of a patent, companies may be less willing to invest millions in AI-driven discovery pipelines.

• AI systems are generative partners.
• Current patent laws require a human inventor.
• DABUS case clarified the definition of an inventor.
• Legal frameworks are anthropocentric.
• A patent encourages investment in innovation.
• The path forward includes modernizing our understanding of contribution.

In practice, the industry is navigating this gray area with a blend of caution and pragmatism. The US Patent and Trademark Office has, at times, issued guidance suggesting applicants need to demonstrate significant human contribution to qualify as a co-inventor with AI. Under different administrations, however, that guidance has shifted toward a “don’t ask, don’t tell” approach, framing AI as merely a sophisticated tool, no different in principle from a calculator. For companies like Insilico, this means keeping humans firmly “in the loop.” As CEO Alex Zhavoronkov explains, human chemists are still essential to synthesize the AI-proposed molecules, create variants, and run preclinical tests. That hands-on work provides the tangible, human contribution required for a patent application.

But the frontier is moving rapidly. What happens when the human role diminishes to simply pushing a button on a fully automated, AI-driven robotic lab? Abbott poses a challenging hypothetical: “What if I asked Claude to cure cancer, and it did? I think it would be inappropriate to claim that I invented that.” This scenario forces us to reconsider the nature of invention itself. Is it the moment of conceptual genesis, or the act of bringing a concept into the physical world? As AI systems grow more capable, the line between assistant and author will only blur further.

The implications extend beyond pharmaceuticals. The US Copyright Office is already refusing copyright for AI-generated images and text, a policy that has raised concerns in creative industries like filmmaking, where tools like OpenAI’s Sora are becoming integral to the production process. The core tension is the same: our legal frameworks for protecting intellectual property are anthropocentric by design. They reward human creativity, sweat, and genius. Machines, no matter how “creative” their output, exist outside this circle of rights and recognition.

The path forward is not about granting personhood to algorithms, but about modernizing our understanding of contribution. The law may need to evolve to recognize new categories of invention or develop sui generis protections for AI-generated outputs that still incentivize investment. For now, the strategy within cutting-edge biotech is clear: document every human step meticulously, ensure a scientist’s hand is visible in the process, and let the AI’s foundational role remain in the press release, not the patent. This dance between technological capability and legal constraint will define the next era of discovery, determining not just who gets the credit, but what gets made at all.