OpenAI just rolled out a new AI model designed specifically for life sciences research. Named GPT-Rosalind, it’s the first in a series aimed at boosting scientific reasoning in fields like biochemistry and genomics, with the goal of speeding up drug discovery and translational medicine.
The model takes its name from British chemist Rosalind Franklin, whose work was crucial in uncovering the double-helix structure of DNA.
Joy Jiao, OpenAI’s head of life sciences research, says GPT-Rosalind is focused on enhancing foundational reasoning in the life sciences. The goal isn’t to replace scientists, but to help them tackle the most time-consuming, analysis-heavy parts of their work.
OpenAI is careful to note that, despite ongoing industry efforts to reduce AI hallucinations, the model is intended for evidence synthesis, hypothesis generation, and assisted analysis only. Final decisions still require human expert judgment and real-world validation.
Here’s the context: In the U.S., it takes an average of 10 to 15 years for a new drug to go from target discovery to approval, and only about one in ten drugs that enter clinical trials actually gets the green light. Globally, roughly 300 million people with rare diseases are waiting for more effective treatments.
GPT-Rosalind comes with enterprise-grade security controls and access management, making it suitable for highly regulated research environments. OpenAI has already partnered with Los Alamos National Laboratory to collaborate on AI-assisted protein and catalyst design.
But real-world challenges remain. No drug fully discovered or designed by AI has passed Phase III clinical trials yet, with only a handful making it into early-stage clinical phases. There’s also the more serious issue of biosecurity risks. According to a previous Axios report, an international team of over a hundred scientists has called for tighter controls on sensitive biological data used to train AI systems, to prevent the technology from being misused to create dangerous pathogens.
To address these concerns, OpenAI is imposing strict access limits. The model is being released as a research preview through its “Trusted Access Program” to select enterprise customers. The first batch includes organizations like Amgen, Moderna, the Allen Institute, and Thermo Fisher Scientific.
Yunyun Wang, OpenAI’s head of life sciences products, says the restricted access is meant to maximize research value while minimizing the risk of misuse. If this approach proves effective for domain-specific models like GPT-Rosalind, expect it to expand quickly to other scientific fields.