For teams working with AI-enabled medical devices, AI medical device post-market surveillance is not only a question of what happens after deployment. It also raises questions about how oversight relates to the setting in which a device is used, who is responsible locally, and how relevant information moves across the health system.
The MHRA’s AI Airlock Phase 3 application places post-market surveillance and lifecycle oversight within that wider coordination challenge. The application call is open, and the GOV.UK page identifies its publication date as 6 October 2026. Its statements describe an area of focus for the call; they should not be read as finalized guidance or as evidence that a particular oversight approach has worked. [1]
What the Phase 3 application brings into focus
The application frames post-market surveillance as a challenge connected to oversight across the lifecycle of AI-enabled medical devices. That framing matters because it directs attention beyond the point of initial deployment: oversight may need to account for how a device is used in different contexts and how people and organisations involved in care share relevant information. [1]
More specifically, the call names three focus areas: accounting for deployment context—including local infrastructure, workflow integration, data availability and quality, and user interaction; adapting surveillance to settings with varying clinical or organisational oversight, such as remote, direct-to-consumer, and at-home use; and communicating information about performance, safety, updates, and rollbacks to regulators, providers, professionals, and patients in a timely way. [1]
For manufacturers, the application makes the data and decision points behind monitoring especially relevant: it says they should be able to access deployment-specific data and have clear criteria for when and how safety or performance trends will be acted on. Care organisations, in turn, are among the settings where local workflows and oversight affect what can be observed and communicated. Phase 3 invites proposals to explore these challenges. [1]
AI medical device post-market surveillance across different settings
A device’s deployment context is one of the coordination themes identified in the application. [1] In practical terms, the same AI-enabled medical device may be used within different organisational and clinical environments. That makes context a useful question for oversight: what information is available locally, and how might people responsible for monitoring understand the device’s use in that setting?
For a manufacturer preparing a proposal, this focus turns into a bounded planning question: identify the deployment-specific data needed to test a hypothesis, then define when and how safety or performance trends would prompt action. A care organisation could use the same proposal to clarify which local workflow and user-interaction details are available to the testing team. [1]
Local oversight is not necessarily uniform
The application also draws attention to variation in local clinical or organisational oversight. [1] This raises a practical coordination question: when responsibilities, workflows, or oversight arrangements differ between settings, how can relevant concerns and observations be understood beyond the immediate team?
For a proposal involving direct-to-consumer or at-home use, manufacturers and care organisations could specify how the testing plan will account for settings where qualified clinical or organisational oversight may be limited. That gives this focus area a distinct role: examining how monitoring is adapted to the visibility available in each use environment. [1]
Communication across the health system
A further theme is communication across the health system about matters such as performance, safety, updates, and rollbacks. [1] The value of this framing is that it treats oversight as involving more than one point of contact: information may need to be understood and shared across the organisations and people connected to a device’s use.
As the call develops, readers can watch for the projects and outputs it describes: bespoke testing plans, public-facing tools or frameworks, and reports setting out candidate-team findings or possible implications for the regulatory framework. These are intended outputs of the programme, not evidence that a particular approach has already improved safety or performance. [1]
What lifecycle oversight means in this framing
Taken together, these themes suggest that the application views post-market surveillance as a continuing concern shaped by deployment, local oversight, and communication. [1] In that bounded sense, lifecycle oversight is not presented solely as a question to settle at the moment of initial deployment.
The application asks proposals to bring enough information to support hypothesis development and testing, including relevant data-access needs. It also describes bespoke testing plans co-developed during onboarding, with each engagement generally scoped to 6–12 months. For manufacturers and care organisations, that makes the call a route to explore how monitoring questions can be tested in a defined setting—not simply a discussion of oversight in the abstract. [1]
What the application does—and does not—establish
The GOV.UK page identifies the Phase 3 application as an open call from the Medicines and Healthcare products Regulatory Agency and gives its publication date. [1] On the supplied evidence, that is the basis for describing the call and its subject.
The application is an open call identifying challenges and goals for proposals, not binding post-market surveillance requirements or finalized MHRA guidance; the page does not establish that any proposed approach has produced successful outcomes. For regulatory and quality teams, it is a prompt to follow authoritative publications and assess applicable requirements through the relevant official sources, not a compliance checklist. [1]
The Phase 3 call puts AI medical device post-market surveillance in the context of deployment conditions, differing oversight environments, and timely information-sharing. Those connected questions give manufacturers and health-system teams a concrete basis for considering where lifecycle coordination needs attention. [1]
;)
