FDA Launches Agency-Wide AI Tool to Aid in Scientific Review

The FDA launched a generative artificial intelligence (AI) tool, dubbed Elsa, designed to help employees such as scientific reviewers and investigators work more efficiently, the agency announced in a press release on Monday.

The tool, which was built within a high-security GovCloud environment, “modernizes agency functions and leverages AI capabilities to better serve the American people,” the release explained. It will allow FDA employees to access internal documents while ensuring information security.

Elsa is already being utilized to speed up clinical protocol reviews, reduce the time it takes to conduct scientific evaluations, and stratify high-priority targets for inspection, according to the press release. | Image Credit: wladimir1804 - stock.adobe.com

“Today marks the dawn of the AI era at the FDA with the release of Elsa,” FDA Chief AI Officer Jeremy Walsh said. “AI is no longer a distant promise but a dynamic force enhancing and optimizing the performance and potential of every employee. As we learn how employees are using the tool, our development team will be able to add capabilities and grow with the needs of employees and the agency.”

Elsa is powered by a large language model, which is a type of generative AI trained on large amounts of data and designed to generate humanlike text or other forms of content based on the data it is trained on.² Large language models can use context to make inferences, answer questions, summarize text, and perform a range of other functions.

The new model was designed to help with reading, writing, and summarizing information to make FDA processes more efficient.¹ Among its functions are summarizing adverse events to assist with safety profile assessments, performing label comparisons quickly, and generating code to help develop databases for nonclinical applications.

“As the tool matures, the agency has plans to integrate more AI in different processes, such as data processing and generative-AI functions, to further support the FDA’s mission,” the press release explained. It also noted that the models are not trained on data submitted by regulated industry, which keeps the sensitive research and data handled by the FDA staff secure.

Elsa is already being utilized to speed up clinical protocol reviews, reduce the time it takes to conduct scientific evaluations, and stratify high-priority targets for inspection, according to the press release. The first step in the FDA’s overall AI journey, Elsa’s introduction came ahead of schedule following an all-center approach. It was a collaboration between leaders and technologists across the agency.

“Following a very successful pilot program with FDA’s scientific reviewers, I set an aggressive timeline to scale AI agency-wide by June 30,” FDA Commissioner Marty Makary, MD, MPH, said in a statement. “Today’s rollout of Elsa is ahead of schedule and under budget, thanks to the collaboration of our in-house experts across the centers.”

References

1. FDA launches agency-wide AI tool to optimize performance for the American people. News release. FDA. June 2, 2025. Accessed June 3, 2025. https://www.fda.gov/news-events/press-announcements/fda-launches-agency-wide-ai-tool-optimize-performance-american-people

2. What are large language models (LLMs)? IBM.com. November 2, 2023. Accessed June 3, 2025. https://www.ibm.com/think/topics/large-language-models