FDA's Postmarket AI Machinery Is Still Under Construction

Two data points, one unfinished picture

Johnson & Johnson’s Monarch bronchoscopy robot, already on the market, just received FDA authorization for new AI features added to the existing platform. FDA also named Cadence as the second participant in its TEMPO digital health pilot, a program built for closer, ongoing agency engagement with software makers.

Both items point toward the same open question: how does FDA handle an AI device that keeps changing after clearance. That is a real and underdiscussed problem. It is not, however, a solved one, and the temptation to read these two events as evidence of a finished postmarket framework overstates what has actually happened.

What the agency itself says

The clearest signal that this machinery is incomplete comes from FDA directly. The agency is actively seeking feedback on generative AI regulations and is still weighing its regulatory approach for genAI-enabled medical devices. A regulator running a public comment process on how to treat a major category of AI systems is not an agency with a settled model. TEMPO, with two named participants, is a pilot in the literal sense: a small, provisional test, not a template with fixed rules. The Monarch authorization shows FDA has a mechanism for incremental AI updates on an already-cleared device. It does not show what that mechanism will look like once generative and adaptive AI models are folded into the same review logic the agency is still debating.

Broader tracking of the SaMD regulatory landscape shows new AI-specific authorization routes appearing across multiple markets, not converging on one design. In the EU, recent life science law updates show MDR, IVDR, and AI Act obligations layering on top of each other in ways that differ meaningfully from FDA’s approach. A device maker planning a global AI roadmap is not choosing between “engage now” and “wait,” because the destination itself is not fixed in any jurisdiction yet.

The real decision, without the thumb on the scale

For compliance and regulatory leaders, the honest framing is a tradeoff, not a verdict. Participating in TEMPO or shaping precedent through incremental submissions does build a working relationship with reviewers, and that relationship has value. It also means committing scarce regulatory affairs bandwidth to a framework FDA has explicitly said it has not finished designing, particularly for generative AI, and early participants absorb the cost of that uncertainty first. Waiting avoids that cost but forfeits influence over how the eventual rules get written, and risks inheriting a framework shaped by others’ priorities.

Neither posture is obviously correct. The right call depends on how far your AI roadmap leans into generative or adaptive capability, where FDA’s position is least settled, versus incremental feature updates on established platforms, where the Monarch precedent already offers a workable path. Treat the current moment as what it is: an agency and an industry still negotiating the rules, not a closing window.

---

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 argument is logically coherent and well-structured: it correctly identifies that FDA’s ongoing public comment processes and pilot status directly contradict any claim of a ‘finished’ framework, an
Source & Claim Verification	Qwen · local	cleared. All factual claims are supported by citations, but some sources are not directly relevant to the claims they are cited for, which could be improved.
Regulatory & Framework Fidelity	Mistral	cleared. The briefing accurately reflects the current state of FDA’s evolving postmarket AI framework and aligns with ISO 42001, EU AI Act, and MDR/IVDR contexts without misrepresenting regulatory maturity.
Technical Accuracy	Llama	cleared. The article accurately reflects the current state of FDA’s regulatory framework for AI devices, highlighting the ongoing development and uncertainty surrounding generative AI regulations.
Bias, Balance & Hype Control	Gemini	cleared. The briefing effectively identifies and counters potential vendor hype by directly addressing the incompleteness of the FDA’s framework and presenting a balanced view of the strategic choices involved
Novelty & Non-Duplication	Grok	held. Synthesizes two same-cycle MedTech Dive wire hits plus ongoing genAI comment process into a familiar ‘FDA postmarket AI still unsettled’ consensus narrative without a clearly non-obvious claim or diff
Validation	DeepSeek	cleared. The briefing’s central claim that the FDA’s postmarket AI framework is unfinished is directly validated by the agency’s own public request for feedback on generative AI regulations.

Sources cited: 14. Validation challenges: 0. Review cost: about $0.04. Learn how these briefings are written and verified.