OpenAI’s closed-access biology LLM entrenches corporate control over life sciences data, deepening extractive AI monopolies
Original framing: “OpenAI starts offering a biology-tuned LLM” — Ars Technica
Scientifically, closed-access AI models like GPT-Rosalind risk limiting the reproducibility and transparency of biological research, as proprietary systems obscure methodological details. Open-access alternatives, such as those developed by the European Bioinformatics Institute, have demonstrated that collaborative models accelerate scientific discovery. Additionally, the lack of peer review in proprietary AI tools raises concerns about bias and error propagation in biological research. The scientific community has long advocated for open data and reproducible methods to ensure robust and equitable progress.
The launch of GPT-Rosalind exemplifies the techno-feudalisation of science, where a Silicon Valley corporation centralises control over biological knowledge under the guise of innovation.
This closed-access model perpetuates historical patterns of enclosure, from colonial biopiracy to the patenting of genetic resources, while sidelining Indigenous and Global South perspectives that have long championed collaborative and communal approaches to knowledge. Scientifically, the proprietary framework risks limiting reproducibility and transparency, undermining the very progress it claims to accelerate. Future pathways must prioritise open-access alternatives, Indigenous data governance, and public funding for equitable infrastructure to prevent a future where a handful of corporations dictate the terms of biological innovation. The systemic insight is clear: without structural intervention, AI-driven science will deepen inequities rather than democratise them, echoing the failures of past enclosure movements.