The PCCP Is the Real Clearance

The clearance date is no longer the question that matters

Buyers evaluating an AI-enabled medical device tend to ask when it was cleared. The more consequential question is what it is allowed to become afterward. FDA’s guidance on predetermined change control plans exists precisely because a model retrained on new data is, functionally, a different device than the one that was authorized, and FDA has built a formal pathway for sponsors to pre-specify how that modification will happen without triggering a fresh submission every time.

This is real regulatory progress on the drift problem. It is also a quiet transfer of scrutiny from the moment of authorization to the fine print of the plan itself. FDA’s broader lifecycle framework for AI-enabled device software functions already commits the agency to evaluating real-world performance after market entry rather than treating clearance as a terminal event, and the defined regulatory framework covering lifecycle risk management is what manufacturers now build their validation and post-market documentation around. The PCCP is where that lifecycle commitment gets operationalized, and it is where the actual risk boundary of the device now sits.

Why this matters more for buyers than for developers

A hospital system, health plan, or diagnostics network procuring an IMDRF Category III or IV software product is not buying a static algorithm. It is buying an algorithm plus a contractual boundary on how much that algorithm can change on its own authority. If the PCCP is narrow, drift is tightly bounded and re-review is frequent. If it is broad, the manufacturer has room to retrain, reweight, or expand indications with far less external checkpoint. Neither is inherently wrong, but treating both as equivalent because both carry an FDA clearance is a due diligence failure.

This gap is widening as regulators globally take divergent approaches to what counts as autonomous enough to require new authorization, and buyers operating across markets face genuine divergence in AI-specific authorisation routes between the US, EU, and other major regulators. A PCCP accepted domestically carries no guarantee of equivalent treatment abroad, particularly as the EU AI Act’s high-risk provisions for embedded medical device AI phase in on their own separate timeline.

FDA has also signaled it is still thinking through how much of this oversight should resemble ongoing clinical judgment rather than a one-time engineering review. The agency’s early exploration of a doctor-like assessment model for AI-enabled devices is explicitly preliminary and not yet guidance, but it confirms the direction: continuous evaluation, not a single approval event, is becoming the operating assumption.

The diligence question to add now

Procurement and compliance teams should stop asking whether a device is cleared and start asking for the PCCP itself: what triggers a modification, what performance bounds constrain it, and what evidence the manufacturer commits to generating before deploying a change. Under an ISO 42001-aligned AI management system, that plan is exactly the kind of change-control artifact an internal audit should be pulling, not taking on faith from a clearance letter.

The clearance tells you where the device started. The PCCP tells you where it is allowed to go.

---

Board record

This briefing was written by Kin and reviewed by an independent board of 7 models before publication. Ruling: CLEARED.

Seat	Reviewer	Finding
Chair · Editorial Judgment	Claude	cleared. The core argument—that PCCPs now define the real risk boundary of AI medical devices—is coherent and defensible, but the piece overstates regulatory consensus by treating FDA’s exploratory ‘doctor-lik
Source & Claim Verification	Qwen · local	cleared. Most claims are well-supported by citations, but a few lines lack direct references, such as the discussion on the widening gap in global regulatory approaches and the specific mention of the EU AI Ac
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately reflects FDA’s PCCP framework and lifecycle approach but understates the EU AI Act’s immediate applicability to medical device AI under MDR/IVDR.
Technical Accuracy	Llama	cleared. The article is generally technically accurate regarding the FDA’s guidance on predetermined change control plans (PCCPs) for AI-enabled medical devices and their implications for regulatory oversight
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and addresses potential vendor hype by shifting focus from a simple ‘cleared’ status to the more nuanced and critical PCCP, providing a strong counter-narrative to
Novelty & Non-Duplication	Grok	held. Punchy buyer-diligence reframe of long-established FDA PCCP/lifecycle AI-device policy, but no original reporting or thesis that clearly clears existing wire and standard medtech regulatory commentary
Validation	DeepSeek	cleared. The central claim that the PCCP defines the device’s operational risk boundary is validated by FDA’s published lifecycle framework and guidance, which explicitly shifts oversight to continuous evaluat

Sources cited: 15. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.