When biotech company Insilico Medicine announced its generative AI platform had "discovered" a promising drug candidate for pulmonary fibrosis, the press release crowned the algorithm as inventor. This framing represents a seismic shift in how companies are publicly characterizing AI's role in pharmaceutical development—moving from tool to autonomous creator. Yet this bold claim masks a regulatory vacuum. Patent offices, drug regulators, and liability frameworks were built for human inventors and corporate entities, not algorithms. If Insilico's AI truly "discovered" the molecule, is the AI the inventor under patent law? Can you patent something invented by a non-human entity? And if clinical trials fail, who bears legal responsibility—the company, the researchers who built the system, or the algorithm itself?
The stakes extend far beyond semantics. Current patent law in most jurisdictions requires human inventors; the U.S. Patent Office has already rejected applications listing only AI as inventor, creating a legal grey zone. Meanwhile, the FDA's drug approval process assumes human accountability at every stage—from initial discovery through post-market surveillance. If a molecule is genuinely "discovered" by AI with minimal human intervention, FDA inspectors face an uncomfortable question: whom do you hold accountable if adverse effects emerge years later? Precedent exists in other fields: AI-generated artwork has triggered copyright disputes, and autonomous vehicle liability sparked years of regulatory debate. But drug discovery differs critically—failure carries life-or-death consequences, not just financial ones.
Remarkably, neither the USPTO, the FDA, nor the European Medicines Agency has issued clear guidance on AI inventorship or accountability. This silence leaves companies in a Wild West of self-regulation. Without explicit standards, some firms may overstate AI contributions to boost investor confidence, while others downplay them to preserve traditional patent structures. The actual solution requires regulators to move beyond abstract questions about machine consciousness and tackle concrete problems: Should patents require a named human "responsible inventor" who certifies the AI's contribution? Should FDA approvals mandate human accountability chains? Should liability insurance explicitly cover AI-discovered drugs? Without these answers, innovation may accelerate but accountability will collapse—and patients in failed trials will pay the price of regulatory indecision.