I’ve seen a few people talking about the FDA’s PCCP guidance lately, so I figured I’d write something quickly on how it relates to cybersecurity, what is still left for prior approval/clearance, and what they’ve expected you to be doing all along.
What a PCCP is
The FDA’s Predetermined Change Control Plan (PCCP) guidance (final for AI software as of August 18, 2025, still a draft for everything else) allows manufacturers to submit a change control plan for certain updates to a device, so they can be pushed out to the field without a new submission. That’s a big win for getting safer, more effective capabilities into the hands of patients and caregivers without undue delay. However, if it’s overly permissive, it could also introduce new risks.
What qualifies – and what doesn’t
The FDA recognized this potential for harm and limited the types of changes that qualify. For instance, swapping in a new wireless card, supporting new operating system versions, or adding new data sources through existing APIs are all covered. Changes that introduce new sources of risk, like significant changes to the software architecture or anything to fix a recall or safety issue, are not. And routine software maintenance and updates, like operating system and third-party component security patches, have been expressly permitted since 2005.
Hopefully that clears things up a bit, though of course grey areas always emerge when going through real-world implementation. If you’re working through one of those grey areas, get in touch and we can go into more depth.