FDA's internal use of AI is poised to reshape drug, device, and biologic regulation. While external debates center on sponsors' AI deployment, the more immediate shift is inside the agency, where AI is expected to advance efficiency, sharpen risk detection, and modernize review operations. As the FDA deputy Chief of Staff remarked: "AI could be the icebreaker to plow through the gridlock."

One major development across FDA Centers is expanded AI‑assisted submission triage. Reviewers are piloting systems capable of scanning thousands of pages to identify content, contradictions, or high‑impact data requiring expert attention. The goal is not to replace scientific review, but to reduce administrative burden. As Commissioner Marty Makary noted, AI can "summarize" large dossiers with remarkable speed. In one pilot, "the AI did in six minutes what it would normally take [a reviewer] two to three days." FDA plans to expand this capability across review teams, especially for drug and biologics programs. Sponsors should expect materials (e.g., meeting packages) to be queried using AI, rather than read front-to-back.

FDA is also deploying AI in compliance. CDRH is strengthening risk‑based inspection targeting by integrating complaint trends, adverse‑event data, and inspection findings to quickly identify early signs of quality drift or manufacturing risk using – especially for software‑driven and AI‑enabled devices – where real‑world performance may subtly deviate over time. For drugs, AI and other tools are being used to proactively surveil and review drug advertising.

AI is equally transformative in FDA's adverse‑event signal identification. Across drugs, devices, and biologics, models can scan FAERS, MDRs, literature, and heterogeneous real‑world datasets to detect safety patterns earlier and more precisely. For drugs and biologics, this means earlier insight into toxicity, demographic risks, or unexpected interactions. For devices, it accelerates detection of performance drift, repeat malfunctions, or user‑interaction anomalies. This shifts post market surveillance from reactive to predictive, enabling faster, targeted interventions.

In addition, FDA is expanding AI into document processing, transparency workflows, and knowledge management, streamlining redaction, accelerating material release, and harmonizing regulatory language across Centers.

As FDA scales its internal AI capabilities, continuous review models and real‑world performance analytics will be more deeply integrated into generation, review, monitoring, and harmonization of scientific evidence across Centers. These advancements signal a future where AI-powered signal detection, high‑fidelity data, and lifecycle‑long integrity anchor FDA's oversight. For sponsors, the opportunity – and challenge – is to modernize systems on pace with FDA's transformation, ensuring readiness as AI‑informed regulation becomes the norm, rather than the horizon.