The FDA has announced the launch of its new generative AI tool named Elsa, aiming to streamline internal operations. While this represents progress in digital transformation, the tool’s deployment also raises important questions about implementation, governance, and long-term effectiveness.

The introduction of Elsa comes on the heels of an internal pilot program completed in early May, which explored AI applications for scientific review, and an ambitious target to scale AI agency-wide by June 30. Launching Elsa ahead of schedule demonstrates the FDA’s focus on integrating cutting-edge technology into its operations. The agency announcement notes that AI assistance will allow the FDA to accelerate protocol reviews, reduce evaluation times, and prioritize inspection targets more effectively. Specific applications include summarizing adverse events, comparing product labels, and generating code for nonclinical purposes. These capabilities are particularly relevant in light of ongoing budget constraints, including a proposed 4% funding reduction in the president’s fiscal year 2026 budget.

“Continued enhancement of the use of AI within the FDA has the opportunity to transform internal operations. While AI cannot replace the expertise of a scientific reviewer or inspector, it can increase efficiency including in generating summaries and organizing data. We encourage sponsors to continue engaging with the FDA on AI’s evolving role in review processes and inspections,” says Dr. Namandjé Bumpus, Strategic Advisor and former Principal Deputy Commissioner of the FDA.

“While the launch of Elsa is a significant step, it’s important to maintain perspective,” says Ashish Bassi, Chief Technology Officer. “The introduction of this tool highlights the FDA’s appetite for integrating new technology, but its impact is yet to be fully realized. AI pilot programs have been in progress for several years, and while they offer promising opportunities, their significance should not be overstated at this stage.”

The FDA emphasized that Elsa operates on a secure platform, GovCloud, ensuring security measures are in place to protect sensitive information. Elsa does not train on data submitted by sponsors, but will be developed based on employee usage, with a goal to enhance its data processing capabilities.

Despite these promising opportunities, some anonymous reports from within the FDA have expressed scepticism about Elsa’s readiness for scientific processes. Critics argue that its current use has been limited to daily administrative tasks and suggest it may be a rebrand of CDER-GPT—a chatbot piloted by the agency last year.

“Introducing new technologies into an already overworked system like the FDA requires careful consideration to maintain the standards that safeguard public health. In a high-risk area such as regulatory review, AI initiatives must be implemented gradually and responsibly, especially given the risk of generative AI producing convincing but incorrect outputs,” said Claudia Federico, Senior Director of AI Transformation.

“The promise of generative AI comes with an equal need for accountability that goes beyond technical deployment,” Federico continued. “It means ensuring that teams are properly trained, fully aware of the limitations, and prepared to critically evaluate every output, as well as strong governance, active oversight, and a culture that prioritizes both innovation and safety.”

Elsa’s launch highlights the growing role of AI in public health and regulatory environments, but its long-term success will depend on more than technological capability. Adoption, training, auditability, and trust will all be critical to achieving sustainable impact.