FDA GenAI public feedbackAll audiences Public feedback on all 26 FDA GenAI questions. Explore each question and the submissions that discuss it.
Assessing clinical risk Q1–Q6
FDA Q1 · The two-axis risk framework Does a two-axis framework, AI device activity and the consequence of relying on an incorrect output, capture the dimensions of risk? 34 of 95 filings name itFDA Q2 · The spectrum of device activity How should the continuum from non-directive to action-directing outputs, and the risk that changes along it, be accounted for? 23 of 95 filings name itFDA Q3 · When an output becomes directive When clinical information goes straight to the patient, does the risk change, and what safeguards help without underestimating patients? 25 of 95 filings name itFDA Q4 · Generalist and specialist users Should it matter whether the clinician using the AI is a generalist or a specialist? 19 of 95 filings name itFDA Q5 · Multi-turn conversations that migrate How is risk assessed when a conversation starts with non-directive information and drifts into action-directing? 22 of 95 filings name itFDA Q6 · Care escalation functions How should under-escalation be weighed against over-escalation? 27 of 95 filings name itDemonstrating clinical competence Q7–Q17
FDA Q7 · The competency-based approach Is the two-step approach, benchmark the AI, then confirm it in clinical use, the right way to evaluate these devices? 23 of 95 filings name itFDA Q8 · Mapping the risk grid to evidence Should an AI device’s position on the risk map help decide how much evidence it must bring before market? 15 of 95 filings name itFDA Q9 · The benchmarking structure Do the ten benchmark competencies, from clinical knowledge to generalizability, add up to enough evidence of safety and effectiveness? 31 of 95 filings name itFDA Q10 · Benchmark contamination and saturation How can a benchmark score be shown to predict real-world behavior? 26 of 95 filings name itFDA Q11 · Clinical confirmation without a prospective trial When can a device be confirmed without a prospective clinical study, and what earns that lighter path? 26 of 95 filings name itFDA Q12 · Statistically meaningful performance How do you get statistically meaningful performance numbers when synthetic inputs are mixed with real ones? 21 of 95 filings name itFDA Q13 · Synthetic data Where is synthetic data good enough, and where is it not? 22 of 95 filings name itFDA Q14 · Comparators and acceptance criteria For open-ended AI outputs, who is the performance comparator: a clinician panel, generalists, specialists, or the human-AI team? 25 of 95 filings name itFDA Q15 · Performance against usual care Could the AI be measured against what would have happened without it: unaided judgment, a delayed specialist, or no intervention? 23 of 95 filings name itFDA Q16 · Independent third parties What role should independent third parties play? 26 of 95 filings name itFDA Q17 · Devices with many functions Does the approach still work for devices built on other model architectures, such as multimodal vision-language models and world models? 12 of 95 filings name itMonitoring after launch Q18–Q24
FDA Q18 · Trading premarket certainty for postmarket monitoring Can greater premarket uncertainty about a GenAI device’s benefit-risk profile be accepted through greater reliance on postmarket monitoring? 29 of 95 filings name itFDA Q19 · Postmarket performance evaluation How should an AI device be monitored after launch, and what sets the cadence? 41 of 95 filings name itFDA Q20 · Machine-based supervisory agents Could AI supervisory agents help carry out postmarket monitoring? 27 of 95 filings name itFDA Q21 · Clinicians, institutions and societies What roles should clinicians and institutions play in monitoring, without diluting manufacturer accountability? 27 of 95 filings name itFDA Q22 · Re-benchmarking after a modification With the premarket competency assessment as the baseline, which post-deployment changes need re-evaluation, and how much? 29 of 95 filings name itFDA Q23 · PCCPs for GenAI devices How can a change-control plan cover changes that cannot be fully specified in advance? 21 of 95 filings name itFDA Q24 · Third-party foundation model changes When the foundation model’s developer changes the model, how does the device maker detect it and respond, so safety and effectiveness are not compromised? 30 of 95 filings name itModels underneath, and agents on top Q25–Q26
FDA Q25 · Foundation Model Master Files Would voluntary Foundation Model Master Files be practical, and useful in premarket review? 22 of 95 filings name itFDA Q26 · Agentic devices What extra oversight does an AI that plans and acts in multiple steps need? 32 of 95 filings name it
Who’s saying what? All audiences shown for comparison.
Provisional mapping: industry includes startups, large companies and consultants; practitioners includes clinicians and health systems. Academia is shown separately. Filings are not a representative population survey.
The conditions matter. Read conditional positions ↗
How much autonomy is accepted? Highest accepted autonomy, compared across two consequence levels.
How to read this
Two debates worth watching. Two themes drawn from the FDA’s questions.
FDA GenAI public feedbackAll audiences The public submissions behind the FDA GenAI discussion. Browse the filings, contributors and original source documents.
Search CoverageAll filings With question-level analysis Without question-level analysis
Explore the 26 FDA questions Counts show coverage among the selected audience, not agreement.
