OpenAI and Los Alamos National Laboratory are developing evaluations to understand how multimodal AI models can be used safely by scientists in laboratory settings.
OpenAI and Los Alamos National Laboratory (LANL) are working together to study how artificial intelligence can be used safely by scientists in laboratory settings to advance bioscientific research. The partnership follows a long tradition of collaboration between the U.S. public sector and private industry to translate innovation into advances in health care and bioscience.
The partnership supports evaluation of frontier-model biological capabilities, including the potential for AI to multiply the speed and impact of science for good. It follows work by organizations using OpenAI technology to support clinical-trial development and evidence-based cancer screening and treatment decisions.
OpenAI and LANL’s Bioscience Division are conducting an evaluation study of how multimodal models such as GPT‑4o can assist people with physical laboratory tasks through vision and voice. The study includes biological-safety evaluations for GPT‑4o and its real-time voice systems, and is intended to contribute to state-of-the-art AI biosecurity research.
The evaluation builds on OpenAI’s early-warning work on biothreat risks and follows its Preparedness Framework. It tests how experts and novices perform and troubleshoot safe, standard laboratory protocols, using tasks as proxies for more complex dual-use concerns.
By examining the uplift in task completion and accuracy enabled by GPT‑4o, the study aims to quantify how frontier models may upskill existing professionals and novices in real-world biological tasks, while understanding and mitigating associated risks.