Helping federal agencies get AI regulation right: Our comment to FDA on generative AI medical devices

At Guidance Global Consulting, our work centers on helping federal health agencies and the health systems they oversee adopt AI responsibly, without losing sight of the patients and providers those systems ultimately serve. That mission is exactly what led us to formally engage with FDA’s Center for Devices and Radiological Health on its discussion paper addressing the regulation of generative AI-enabled medical devices, filed jointly with Profound Ventures under docket FDA-2026-N-7874.

FDA’s paper addresses a question with no established precedent: how should the agency evaluate a medical device whose behavior is not fixed, one that can respond differently to similar inputs, may be built on a foundation model the manufacturer does not fully control, and can continue to evolve after it reaches a hospital or clinic. FDA has been clear that this paper is meant to inform, not finalize, its regulatory approach. For a firm built to help agencies and health systems navigate exactly this kind of AI adoption and regulatory strategy work, contributing to that process while it is still being shaped was a natural extension of what we do every day for our clients.

I co-authored this comment with Brian Meshkin, Managing Partner of Profound Ventures. Our collaboration brought together two distinct bodies of experience: his in building and scaling diagnostics and AI-enabled healthcare ventures, and mine in 12 years at CMS overseeing Medicare beneficiary advocacy and innovation, alongside GGC’s ongoing work advising health systems and regulatory agencies on AI adoption and quality measurement.

A summary of our core recommendations, and the thinking behind them, reflects the same priorities we bring to every engagement:

Clinical validity and clinical utility should be evaluated as separate evidentiary questions. A device can perform accurately without demonstrably improving a clinical outcome or decision. We recommended FDA require sponsors to establish both, function by function, rather than allowing strong performance on one to imply the other. This is the same distinction we help clients build into their own quality and evidence frameworks.

Regulatory burden should account for deployment capacity, not only development cost. Much of the current discourse around AI device oversight, understandably, centers on what it costs a company to bring a device to market. We raised a related and, in our view, underexamined concern: a well-benchmarked device deployed in a rural hospital or an under-resourced clinic does not come with the staffing capacity to monitor it the way the proposed framework assumes. This is a gap we see routinely in our own work with under-resourced health systems, and a genuinely risk-proportionate approach has to account for it.

Front-line clinicians should have a formal role in device evaluation. A device can score well on every accuracy benchmark and still underperform in clinical use due to workflow misalignment, alert fatigue, or miscalibrated trust over time. We recommended FDA incorporate human factors testing with practicing clinicians as a standing element of premarket evaluation, consistent with our own belief that clinician input belongs at the design table, not just at the point of adoption.

FDA can extend existing regulatory infrastructure rather than build new mechanisms from scratch. We pointed to precedent in laboratory developed test oversight, established post market monitoring pathways, and CMS’s own experience evaluating AI tools that extend clinical roles, infrastructure FDA can draw on directly.

This kind of engagement, bringing operational and regulatory experience directly into an open federal rulemaking process, is at the heart of what GGC exists to do. We were glad to contribute this comment while FDA’s framework is still being developed, and we look forward to continued engagement with the agency as this process moves forward.

The public comment is available for review under docket FDA-2026-N-7874 at regulations.gov.

https://www.regulations.gov/comment/FDA-2026-N-7874-0094

Anita Monteiro, Co-Founder & CEO, Guidance Global Consulting

Brian Meshkin, Chairman & CEO, Profound Ventures