FDA seeks input on regulatory framework for generative AI-enabled medical devices

The FDA published a discussion paper on August 18, 2026 outlining preliminary considerations for regulating generative AI (GenAI)-enabled medical devices, opening a formal public comment period through October 19, 2026. The paper was issued by the agency's Digital Health Center of Excellence (DHCoE), housed within the Center for Devices and Radiological Health (CDRH). The FDA described the effort as a foundational step toward a purpose-built regulatory framework for this device category, aligned with a stated administration priority to accelerate AI-enabled medical product development while preserving patient safety.

---

What it covers

The discussion paper proposes a two-axis risk assessment framework to stratify GenAI-enabled devices and calibrate regulatory expectations accordingly. For premarket evaluation, the FDA advances a competency assessment model — drawing conceptual parallels to physician credentialing — that would combine non-clinical benchmarking with clinical confirmation to establish that a device performs as intended before reaching patients. On the postmarket side, the paper outlines multiple risk-proportionate monitoring approaches, acknowledging that GenAI systems may behave differently over time than static software.

The paper also addresses two technically complex areas: foundation models, which may underpin multiple downstream applications, and agentic AI systems capable of autonomous decision-making with limited human oversight. For each domain, the FDA poses targeted questions to guide stakeholder input and inform the eventual framework.

---

Why it matters

The paper represents one of the FDA's most substantive steps toward developing regulatory expectations specifically tailored to GenAI-enabled medical devices, which the agency says may pose unique risks compared with traditional software and other AI-enabled devices. The competency-based premarket evaluation concept, if adopted, would introduce a structured, multi-stage evidentiary standard that goes beyond current performance benchmarking norms, with direct implications for development timelines and submission strategies.

The treatment of foundation models raises unresolved questions about how regulatory responsibility will be allocated across the AI supply chain — among model developers, device integrators, and clinical deployers. Agentic AI systems present novel safety, oversight, and postmarket monitoring challenges that existing surveillance mechanisms were not designed to address. The FDA's framing of this framework as a potential "model for regulators around the world," in the words of CDRH Director Michelle Tarver, signals that the approach developed here may influence international regulatory alignment.

---

What to watch

• Comment period deadline — October 19, 2026: Stakeholders including device manufacturers, AI platform developers, clinical institutions, and patient advocates can submit feedback under docket FDA-2026-N-7874 on Regulations.gov. Positions articulated here are likely to shape any subsequent binding guidance or rulemaking.
• Foundation model accountability: The paper's treatment of foundation models remains conceptual. Watch for whether the FDA moves toward holding upstream model developers directly accountable or places regulatory obligations primarily at the device-integrator level — a distinction with significant implications for AI infrastructure companies entering healthcare.
• Reconciliation with existing AI/ML frameworks: Monitor how this GenAI-specific framework will be aligned with the FDA's prior work on AI/ML-based Software as a Medical Device (SaMD), including the Predetermined Change Control Plan concept, to avoid regulatory fragmentation across device categories.
• International coordination: Given the FDA's stated aspiration for global regulatory leadership, track whether the International Medical Device Regulators Forum (IMDRF) or other multilateral bodies initiate parallel workstreams — and whether harmonization or divergence emerges as the dominant dynamic.

---

Access the AllSci platform to explore the science behind the news.

---

Spot something wrong? Report an issue with this article