FDA Upgrades Internal AI Tool
The U.S. Food and Drug Administration (FDA) has expanded its AI capabilities as part of a broader modernization effort with the launch of Elsa 4.0, an upgraded internal generative AI tool designed to improve agency efficiency.
Elsa 4.0 now includes custom agents and can generate documents, analyze and visualize data, search the web and large document repositories, convert documents and images into searchable text, and perform voice-to-text dictation.
In addition, the agency has consolidated disparate FDA data sources, systems, and portals into a single platform called Harmonized AI & Lifecycle Operations for Data (HALO). HALO has been integrated with Elsa to support data querying and future AI developments.
“Elsa’s new capabilities once again position FDA as a leader in deploying AI tools that empower staff,” said Marty Makary, MD, MPH, the outgoing FDA Commissioner. “Removing tedious burdens for staff enables them to focus more on science and makes their work streams more efficient and enjoyable. We have some of the best scientists in the world, and we need to take good care of them.” Dr. Makary said during a meeting that about 90% of FDA employees are now using Elsa.
“With the consolidation of our application and submission data sources, systems and portals into HALO and the improvements in Elsa’s capabilities, Elsa will soon become the main entrée into the FDA’s systems and data. Previously, FDA staff would bring data to Elsa. Now, Elsa sits on top of our data,” said Jeremy Walsh, Chief AI Officer. “Integrating AI into our workflows is an urgent priority that will allow us to rapidly advance regulatory science and deliver more cures and meaningful treatments to patients faster.”
The AI system is built within a FedRAMP High secure Google Cloud Platform environment designed to safeguard sensitive agency data, with human subject-matter experts involved at every stage to verify inputs and outputs. Although the system is not connected to the internet, it can still access refreshed secure web data to generate responses.
In February 2026, the agency transitioned from Anthropic to Google Gemini as its primary AI platform.
The first iteration of Elsa launched in June 2025, with additional enhancements introduced since then to support agency operations. The model was designed to accelerate clinical protocol reviews and strengthen regulatory rigor as part of the agency’s broader effort to shorten review and evaluation timelines.
ASCO AI in Oncology is published by Conexiant under a license arrangement with the American Society of Clinical Oncology, Inc. (ASCO®). The ideas and opinions expressed in ASCO AI in Oncology do not necessarily reflect those of Conexiant or ASCO. For more information, see Policies.