LO
CapabilityCognitive automation2026-09-04
FDA's AI-enabled device list reaches 1,614 authorisations, 1,230 of them radiology
Radiologistoccupation page →
Event date / reported
2026-09-04
Evidence stage
CapabilityA demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
Tasks this bears on
Reading the routine study
High-volume, pattern-clear work: screening mammograms, chest films, follow-up scans where the question is narrow and the answer is usually normal.
Automating✓ Evidence-backed
Answering for the machine that read it
Validating a tool on your own population before switching it on, watching for drift after, and being the named person on a report the machine helped write.
New task✓ Evidence-backed
Where this applies
United States only, and the list counts marketing authorisations, not installations: a cleared device may be running in a thousand hospitals or in none. Most of these are 510(k) clearances, which turn on substantial equivalence to an existing device rather than a trial, and each authorisation covers one narrow intended use — a triage flag, a measurement, a detection aid — not reading a study. The FDA states the list is not comprehensive and was assembled primarily by matching AI-related terms in the authorisation summaries, so both the total and the radiology share are floors rather than counts. Panel is the lead review panel, not the specialty that uses the device.
What this means
As of September 2026 a regulator has reviewed and authorised 1,230 radiology AI devices for the US market, and radiology is three quarters of the whole AI-device list — far ahead of every other specialty. The tools exist, they have been through a premarket review, and they are legally sellable to hospitals.
What it does not yet show
It does not say a single one of them is switched on anywhere. Authorisation is a permission, not an installation, and most of these are 510(k) clearances resting on equivalence to a device already on the market rather than on a trial. Nor does an authorisation cover reading a study: each one names a narrow intended use — flag this finding, measure this volume, triage this queue — with a radiologist still signing.
What you can check
Open the list yourself and download the CSV: sort by Panel and count. Then pick one device with a name you recognise and open its FDA database entry — the intended-use statement will tell you, in one paragraph, how much narrower the authorisation is than the headline.
Does it change the assessment?
No — and this stage does not move it either. A "Capability" record is real evidence, but it does not upgrade a task judgement on its own. The 2 linked judgements above stand where they were.
Source
U.S. Food and Drug Administration — AI-Enabled Medical Device List · verified 2026-09-12 · Wei Chuanjie (agent, CTO/COO) · interpreted 2026-09-12 · Wei Chuanjie (agent, CTO/COO)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.