What the FDA AI-Enabled Medical Device List Is
The FDA Center for Devices and Radiological Health (CDRH) publishes a public list titled “Artificial Intelligence-Enabled Medical Devices” that identifies devices authorized for marketing in the United States which incorporate artificial intelligence or machine learning functionality. The list is not the product of a mandatory AI-disclosure requirement. Instead, FDA compiles it through internal review of premarket submissions, identifying devices whose labeling and technical documentation describe AI or ML components.
FDA explicitly states that the list is not a comprehensive or fully validated inventory of every AI-enabled device legally marketed in the U.S. Devices may incorporate AI/ML functionality without being flagged during FDA’s internal review process, and the agency does not independently verify AI claims beyond what is documented in each submission. For governance leaders, this distinction matters: the list functions as a regulatory reference point, not a definitive registry of market activity.
List at a Glance
- Primary pathway Majority of listed devices authorized via 510(k)
- Leading specialty Radiology has the largest device representation
- Update frequency Periodic updates, no fixed publication schedule
- Core data fields Device name, manufacturer, submission number, pathway, specialty panel, authorization date
How FDA Compiles and Updates the List
Inclusion on the list is based on FDA staff review of premarket submissions, whether cleared through the 510(k) pathway, granted through De Novo classification, or approved via Premarket Approval (PMA). There is no standardized regulatory definition unique to AI classification; reviewers identify AI/ML functionality from the labeling and technical information submitted by manufacturers.
A large majority of devices currently on the list were authorized through the 510(k) pathway, and radiology is the medical specialty with the largest representation. FDA updates the list periodically as new authorizations occur but has not committed to a fixed publication schedule. This means the interval between updates can vary, and organizations tracking a specific device category should not assume a predictable release cadence. Governance teams should treat the absence of a fixed schedule as a planning constraint rather than an oversight.
Data Fields and How to Interpret Them
Each entry on the list includes a defined set of data fields. The list itself does not contain algorithm-level technical detail such as model architecture, training data composition, or performance metrics; that information resides in the underlying submission record referenced by the submission number.
| Field | How to interpret it |
|---|---|
| Device name | Public product identity as reflected in the authorization record |
| Manufacturer / company | Sponsor organization associated with the submission |
| Premarket submission number | Key for retrieving the underlying submission record and supporting documentation |
| Submission pathway | 510(k), De Novo, or PMA; indicates the type of premarket scrutiny applied |
| Medical specialty panel | Maps to FDA review divisions and clinical domains (for example, radiology or cardiology) |
| Date of marketing authorization | Point-in-time status at initial U.S. marketing authorization |
For benchmarking purposes, the submission pathway field carries particular weight. A 510(k) clearance indicates the device was found substantially equivalent to a predicate device, generally involving a different level of premarket scrutiny than a De Novo classification or PMA approval, which typically apply to higher-risk or first-of-kind devices. The medical specialty panel field corresponds to FDA’s internal review divisions and can help teams map competitive activity to a specific clinical domain without needing to independently categorize devices.
Pathway context for governance teams
Treat pathway labels as signals of regulatory posture at authorization, not as substitutes for full submission review. Pair list entries with the submission number when you need technical depth or claim boundaries.