To guard against both synthetic and natural biological threats, OpenAI announced the Rosalind biodefense program on May 29, 2026. It provides approved developers and government partners with access to GPT-Rosalind, a frontier reasoning model named after British chemist Rosalind Franklin and focused on life sciences.
Given the model's power in molecular biology, genetic engineering, and disease transmission reasoning, OpenAI has imposed strict access controls to prevent biosecurity risks and misuse.
The program has two tracks. The developer track is for nonprofits, academic institutions, and small-to-medium teams, covering projects like epidemiological modeling, early pathogen detection, genetic screening, and non-pharmaceutical interventions. The government track is for the U.S. government and allied partners, supporting high-impact work such as outbreak response planning, medical countermeasure development, and early warning systems. To lower the barrier to entry, OpenAI will directly cover the model's operating costs.
Early partners include several top research institutions and industry alliances. Lawrence Livermore National Laboratory is using the model to evaluate medical defense countermeasures. Johns Hopkins University Applied Physics Laboratory has integrated it into a protein engineering platform for therapeutic screening. The Coalition for Epidemic Preparedness Innovations (CEPI) has adopted the model to accelerate its 100-day vaccine development mission. On the application security side, teams like SecureDNA and Fourth Eon have brought the model into DNA screening and biosecurity defense construction.