Practical implication
1) Inventory AI use cases that could feed regulated decisions (including PV/safety-related analyses or summaries used in regulatory interactions) and map owners/vendors. 2) For each AI use case, document context of use, intended outputs, and controls (e.g., review/approval, traceability, change management) aligned to a draft-guidance tracking plan. 3) Track the Federal Register notice and plan whether to submit comments within the window described by FDA (90 days after Federal Register publication).
FDA has posted a January 2025 draft guidance titled “Considerations for the Use of Artificial Intelligence To Support Regulatory Decision-Making for Drug and Biological Products.” For pharmacovigilance (PV), quality, and PV technology/computer system validation (CSV) teams, the immediate impact is governance and inspection/submission readiness: organizations should be able to identify where AI is used (internally or through vendors) to generate, transform, summarize, or interpret information that could be used in regulated decision-making.
On FDA’s guidance landing page, the document is characterized as a Draft, Level 1 guidance and explicitly labeled “Not for implementation,” with the page also noting it contains non-binding recommendations. The same landing page lists Docket Number FDA-2024-D-4689 and links to a Federal Register notice.
The draft PDF further states that comments and suggestions should be submitted within 90 days of publication in the Federal Register notice announcing the draft document. The fact sheet does not provide the Federal Register notice date or content, so PV and regulatory intelligence teams should monitor docket activity to confirm timing.
While the fact sheet does not provide details from inside the body of the draft guidance, the status and framing alone can trigger internal readiness work for AI-enabled processes that touch regulated outputs. PV teams may wish to focus on three practical areas aligned with the draft’s positioning: (1) inventory and map AI-enabled workflows that could feed regulated decisions (including safety-related analyses or summaries used in regulatory interactions), covering both internal tools and vendor-provided systems; (2) for each AI use case, document the context of use and intended outputs and confirm governance controls are defined and retrievable (e.g., review/approval, traceability, and change management) aligned to a plan for tracking the draft guidance; and (3) track the docket and the Federal Register notice referenced by FDA and consider whether to submit comments within the 90-day window described in the draft PDF.
The organizations most exposed are those using AI to support regulatory decision-making for drug and biological products, including sponsors/MAHs and their vendors. Within PV, the roles most likely to be involved are PV Technology / CSV (system ownership and validation posture), PV Quality (governance and inspection readiness), and Regulatory Intelligence (docket monitoring and comment planning), with escalation to the QPPV and PV Operations leadership as needed.
This document is a draft and “Not for implementation,” but it is also a Level 1 guidance, which can signal broad applicability. PV, quality, and regulatory intelligence teams should use this posting to trigger a targeted review of AI use cases that could influence regulated decisions, and align documentation, controls, and monitoring plans to FDA’s evolving expectations.
