The FDA is migrating its internal AI tool Elsa from Claude to Gemini. Learn how this shift may affect regulatory reviews, submission strategy, and FDA timelines.
For companies preparing medical device or pharmaceutical regulatory submissions to the FDA, these developments reinforce the importance of clear documentation and proactive regulatory planning.
What Is Elsa? The FDA’s Internal AI Review Tool
In June 2025, the FDA launched Elsa to help FDA reviewers analyze large volumes of regulatory documentation. Developed by Deloitte and built on Anthropic’s Claude model, the system was deployed within a secure AWS GovCloud environment.
Elsa was designed to help reviewers:
- summarize adverse event reports
- compare product labeling
- review clinical protocols
- identify potential inspection priorities
The initiative was introduced during a period of increased workload at the agency. Since early 2025, the FDA has lost more than 4,300 employees, leaving the organization roughly 20% understaffed. AI tools like Elsa were intended to help remaining reviewers manage growing regulatory workloads.
Why the FDA Is Migrating Elsa to a New AI Model
On February 27, 2026, a federal directive required agencies to discontinue use of Anthropic’s Claude model. The FDA subsequently began migrating Elsa to Google’s Gemini model.
Although both systems are large language models, they operate differently. Reviewers must adapt prompting approaches and validate outputs again – particularly for scientific and regulatory tasks where accuracy is essential.
This transition is occurring while the FDA is already operating with reduced staffing levels, creating additional operational challenges for the regulatory review process.
Implications for FDA Regulatory Submissions
For companies preparing regulatory submissions, the current environment places increased emphasis on submission quality and preparation.
Reviewer bandwidth is constrained, and AI-assisted tools used in the review process are evolving. This may limit the ability for extended back-and-forth clarification during regulatory review.
Companies that submit well-structured, clearly organized regulatory documentation are likely to be better positioned than those relying on iterative revisions during the review process.
How Regulatory Teams Should Respond
Regulatory teams may want to consider several adjustments:
- Prioritize submission readiness. Well-prepared submissions entering the queue now may move forward sooner than those delayed into a growing backlog.
- Focus on clear documentation. Structured, consistent submissions help reviewers efficiently navigate complex information.
- Engage the Q-Submission process early. Pre-submission interactions remain one of the most effective ways to identify potential issues before formal review.
- Maintain inspection readiness. As AI tools influence inspection prioritization signals, maintaining strong, audit-ready quality systems remains essential.
Key Takeaways for FDA Submission Strategy
The FDA is currently navigating several overlapping transitions, including staffing reductions, leadership changes, and the migration of an internal AI review platform.
Together, these developments may affect how regulatory submissions are reviewed and processed.
For sponsors preparing submissions, clarity, preparation, and proactive regulatory engagement remain critical to successful FDA interactions.
At AlvaMed, we support medical device and life sciences companies with regulatory strategy, submission preparation, and quality system development to help navigate evolving FDA expectations.
To learn more, explore our services for AI in regulatory and quality workflows, including:
AI coaching and facilitation for regulated medical device teams
-
AI transformation and implementation for regulatory and quality systems
