When Insilico Medicine announced in 2021 that its generative AI platform had 'discovered' a novel drug candidate for pulmonary fibrosis, the company's enthusiastic press release highlighted a milestone that seemed to vindicate the promises of computational drug development. The AI system had screened millions of molecular combinations and identified a compound that showed promise in preclinical models. However, beneath this headline-grabbing claim lies a more complicated reality: the AI proposed the molecule structure, but human chemists synthesized it, human biologists conducted experiments, and human physicians interpreted the results. Insilico's framing—attributing primary discovery credit to the AI platform—glossed over these human contributions and raises urgent questions about how the biotech and regulatory industries should formally credit innovations in an age of AI-assisted research.

The attribution challenge extends beyond marketing claims into intellectual property law and scientific publishing norms. The World Intellectual Property Organization (WIPO) has received multiple patent applications listing AI systems as inventors, forcing patent offices worldwide to grapple with whether algorithms can legally qualify as inventors under existing frameworks. Most jurisdictions, including the U.S. Patent and Trademark Office, currently require patents to list human inventors, though WIPO's 2022 conversation paper acknowledged this may need revision. Meanwhile, scientific journals have begun revising authorship guidelines. The International Committee of Medical Journal Editors (ICMJE) updated its standards to permit AI as a research tool but explicitly prohibit it from qualifying as an author—a reflection that while AI contributes intellectually, human accountability remains essential for scientific integrity. These institutional responses suggest emerging consensus that AI should be credited as a tool rather than an autonomous discoverer, yet no unified standard exists globally.

Policymakers are beginning to address this gap. The European Union's proposed AI Act contains provisions addressing algorithmic accountability in high-risk applications, including pharmaceuticals, though it remains ambiguous about attribution specifically. Several biotech companies, including Deep Genomics and Exscientia, have adopted more measured language in recent announcements, crediting their 'AI-assisted' discovery pipelines rather than claiming AI as primary discoverer—a subtle but important distinction. However, without formal regulatory guidance from the FDA or EMA on how to document AI's role in drug development for approval submissions, companies lack clear incentives to standardize attribution practices. Industry observers argue that clearer frameworks are needed: explicit guidelines distinguishing between AI roles in hypothesis generation, compound screening, and validation, coupled with standardized documentation in regulatory filings. Until such frameworks exist, the tension between AI's genuine contributions and human expertise will continue generating both scientific disputes and potential legal complications in one of technology's most consequential applications.