News

FDA launches “Elsa” AI tool to aid drug approvals

12 June 2025

Artificial Intelligence (AI),machine learning with data mining technology on virtual dachboard.Double Exposure,Businessman hand working concept. Documents finance graphic chart

The U.S. Food and Drug Administration (FDA) has introduced “Elsa,” a generative AI tool designed to enhance the efficiency of its regulatory operations. Elsa's deployment comes nearly a month ahead of the agency’s original schedule, following what was reported as a successful pilot program with scientific reviewers. FDA Commissioner Dr. Marty Makary highlighted Elsa's potential to transform agency workflows, stating that tasks previously taking days can now be completed in minutes. The tool is part of FDA's broader initiative to integrate AI across its processes, aiming to improve operational efficiency and better serve public health. But while officials are optimistic about its potential, there are questions and concerns about its intended uses, its capabilities, and the existence of guardrails.

There are important questions about the purposes for which Elsa will be used. For one, it is unclear whether Elsa is chiefly a tool for administrative efficiency or whether it is intended for more substantive scientific review. Elsa is reported to be a generative AI tool based on a large language model platform. In May, FDA announced Elsa’s development and proposed use, suggesting that it is intended to streamline, in particular, the time-consuming process of drug evaluation and regulatory review. Its roll-out is agency-wide and it appears to be available for all FDA staff, not merely review staff.

FDA’s announcement states that it is designed to assist employees in repetitive tasks such as reading, writing, and summarizing. At the same time, its announced capabilities include summarizing adverse events to support drug safety assessments, performing expedited label comparisons, aiding in the identification of high-priority inspection targets, generating code for nonclinical database development, and clinical trial protocol review.

One key point underscored in Commissioner Makary’s announcement is that Elsa is not being trained on data submitted by industry. Yet, it is unclear which data the AI will be able to access, and whether suitable guardrails are in force. For example, if Elsa is used to analyze data submitted as part of a new drug application (NDA) or biologics license application (BLA), it remains unclear whether the AI be required to “forget” this data when it moves on to another sponsor’s NDA or BLA. The statutory authority for sponsors to rely on data they do not own or for which they do not have a right of reference is limited, and there are limits to FDA’s authority to access and allow such data to inform its review of a subsequent marketing application.

As with other generative AI systems, Elsa’s performance may be inconsistent, and will require careful oversight, particularly if it is involved in generating information upon which regulatory decisions are based. There are reports of FDA staff expressing concerns regarding the tool's readiness and accuracy. The balance between innovation, accountability, and transparency is vital to industry and other stakeholders. As FDA further embraces AI, we will continue to monitor these developments closely for information on the platform’s capabilities and for signs of its use in the application review process.

Authored by Jason Conaty.