The FDA 510(k) AI medical device clearance list, explained

FDA maintains a public list of AI/ML-enabled medical devices that have received clearance. As of mid-2026, that list covers approximately 950 devices. It's the most comprehensive public record of what FDA has approved in the AI device space — and it's a spreadsheet.

What FDA's AI device list is

The official name is the "Artificial Intelligence and Machine Learning (AI/ML)-Enabled Medical Devices" list. FDA publishes it on its website and updates it periodically — not in real time. The list captures devices that received 510(k) clearance or De Novo authorization where the device incorporates AI or ML as a core component of its function.

The list is notably not exhaustive of all software-based medical devices, only those where FDA explicitly categorized the device as AI/ML-enabled. Devices cleared under general software product codes that incorporate AI functionality but weren't specifically flagged may not appear.

What each field means

Device Name is a free-text field completed by the applicant. It's not standardized. The same type of device may appear under a dozen different names across applicants. Device name is useful for understanding what a specific company cleared, but not reliable for searching by device type.

510(k) Number is the structured identifier for the clearance, in the format K followed by six digits. This is the field to use for cross-referencing against the full 510(k) database, pulling summary documents, and tracing predicate relationships. Every analysis should anchor to K-numbers, not device names.

Device Type / Product Code identifies the regulatory category. For AI devices, a handful of product codes cover most clearances: QMF for general AI diagnostics, OZO for radiology AI, QNC for pulmonary AI, QLL for musculoskeletal AI.

Algorithm Type is the most AI-specific field. FDA categorizes devices as diagnostic, triage, treatment planning, monitoring, or general. This categorization matters for regulatory strategy: devices in the same indication but different algorithm type categories may face different performance documentation expectations.

Trends visible in the list

Radiology dominates. Roughly 60% of cleared AI devices are radiology applications — CT analysis, chest X-ray triage, MRI interpretation, mammography screening.

Clearance volume is accelerating. The number of AI device clearances per year has grown roughly threefold since 2020. FDA's Digital Health Center of Excellence has processed a significant backlog and is now handling new submissions at higher volume.

A small number of companies account for most clearances. In most indications, two or three companies have cleared multiple devices. These repeat clearers have established predicates and, in some cases, PCCP approvals that allow algorithm updates without new submissions.

Why the list falls short for serious research

The static spreadsheet format creates three practical problems: no predicate relationships, no link to enforcement history, and no alert mechanism. If a competitor cleared a device last week in your indication, you won't know from the list until you check manually.