Skovos comment to FDA on generative AI in medical devices
In September 2026, Skovos filed a public comment on the FDA discussion paper on generative AI-enabled medical devices (docket FDA-2026-N-7874). Our position: a generative AI device is safe in use only where the hospital can list it, limit it, log it, and stop it.
List it. Limit it. Log it. Stop it.
Our comment responds to Questions 18 through 22, 24, and 26 on postmarket monitoring, change control, and agentic devices, from the deploying institution's point of view.
- List it: an inventory of the AI functions in use, their production versions, accountable owners, and authorized clinical settings.
- Limit it: written permissions for what a device may read, write, order, or send, checked when an action occurs.
- Log it: a complete, tamper-evident record of inputs, outputs, and actions in the institution's custody.
- Stop it: an institution-controlled way to suspend or revoke the device's ability to act.
We recommend that FDA name these capabilities as expectations for postmarket monitoring and require manufacturers to make them possible for deployers.
Read the public comment
Read the full Skovos comment (PDF, 3 pages)
View the public filing on Regulations.gov. Comment ID: FDA-2026-N-7874-0130.
This is Skovos's response to an FDA discussion paper and request for feedback. It is not FDA guidance, clearance, approval, or endorsement of Skovos.