The Food and Drug Administration said it launched Elsa 4.0, a major upgrade to its internal AI assistant, and completed the consolidation of more than 40 applications and submission data sources into a unified platform called HALO. The integration is designed to let FDA staff query agency data and build workflows without manual document uploads, a shift officials say will accelerate reviews and regulatory science while easing administrative burden on researchers and inspectors.
Commissioner Marty Makary said the tools aim to free staff to focus on scientific work, while Chief AI Officer Jeremy Walsh called embedding AI into workflows an urgent priority to speed patient access to treatments. New features in Elsa 4.0 include custom agents, document generation, quantitative analysis and visualization, OCR, secure web search, dictation, and enhanced search across large repositories.
The agency said Elsa runs in a FedRAMP High Google Cloud environment, does not train on staff inputs or industry submissions, and is not internet-connected; human experts remain in the loop to validate outputs. The effort, following Elsa’s 2025 debut, underscores regulators’ broader push to modernize operations with AI under tightening federal governance and security standards.
Related articles:
NIST AI Risk Management Framework 1.0
Google Cloud FedRAMP Authorization Overview
