FDA Leadership Shake-Up Puts Its AI Strategy Under Pressure
Makary drove the FDA’s centralized AI push, including Elsa and plans to use agentic AI in reviews, inspections, and administrative work. However, his departure—along with AI chief Jeremy Walsh and acting CIO Sridhar Mantha—has left the program without clear central leadership. As a result, the FDA could continue its agency-wide strategy or return to fragmented, center-led AI adoption.
Elsa Remains Active, but the Governance Structure Is Unclear
Despite the leadership turnover, acting commissioner Kyle Diamantas has said that artificial intelligence remains a major FDA priority. Therefore, the agency does not appear ready to abandon Elsa or its broader AI modernization plans.
The FDA developed Elsa from CDER GPT, an earlier tool created within the Center for Drug Evaluation and Research. The agency then added retrieval-augmented generation to connect the model with controlled and trusted information sources.
As a result, FDA employees can use Elsa to summarize public comments, review long regulatory histories, and organize information across complex submissions. Moreover, these tools can reduce repetitive workloads. Still, available information suggests that FDA staff retain responsibility for final regulatory decisions.
Pharma Sponsors Still Cannot See How FDA AI Reviews Their Data
The FDA has not changed its rules for sponsor use of AI. However, pharma companies still cannot see how reviewers use AI to search, summarize, or compare submission data. This lack of transparency increases the risk that poor metadata, inconsistent labels, and fragmented records could weaken an otherwise strong regulatory package.
Validation Guidance May Fall Behind the Speed of AI Adoption
Leadership changes may also slow the development of formal AI guidance. The pharmaceutical industry still needs clearer expectations for validating AI tools used in clinical trials, quality systems, and other regulated activities.
Meanwhile, the FDA appears to be returning to its traditional guidance and policy development process. This approach creates more reliable regulatory expectations. Yet it also requires time, consultation, review, and formal publication.
AI technologies can change significantly within one year. Therefore, even a relatively fast FDA guidance process may struggle to match the speed of AI adoption across pharmaceutical operations.
What AI-Assisted FDA Reviews Mean for Pharma Submissions
Pharma companies cannot control the FDA’s AI strategy, but they can control submission quality. Clear metadata, consistent data labels, traceable evidence, and well-organized regulatory histories will help reviewers and their AI tools find and interpret information accurately. As FDA AI use expands, strong data integrity and human oversight will become essential for regulatory readiness.
Zamann Pharma’s Digital Solutions for GMP-Regulated Operations supports compliant digital workflows, master data control, computerized systems validation, and safe AI use in quality processes. Explore the service to strengthen data integrity and prepare regulated systems for a more AI-assisted FDA environment.
Source: Pharmavoice.Com