← All 95 filings
Saurabh Sharma
Public / patientsMember of the publicFiled September 14, 2026392 wordsFDA-2026-N-7874-0096
What they argued
RecovryAI’s one-line reading of the filing.
M5 is the whole filing (Q24): devices built on third-party foundation models can be kept under control, conditioned on local small-language-model deployment or private locked environments where clinical utility permits, strict multi-version lifecycle management with backward-compatibility matrices and deprecation schedules in the QMS, and, where hosted APIs are unavoidable, golden synthetic benchmark suites on a fixed cadence, semantic drift analysis and guardrail telemetry; he asks FDA to recognise localized version locking as an acceptable change-control mitigation under PMA or 510(k) modifications. Nothing is said about autonomy, evidence proportionality, the premarket-postmarket trade, or competency benchmarking. Type: no professional role, organisation or credential is stated anywhere in the comment and it was filed under the Individual Consumer category.
Themes it raises
3 of the 21 themes in the docket.
Watching the device after it shipsFDA Q19, Q20
Controlling a device that keeps changingFDA Q22, Q23, Q24, Q25
How this fits rules that already existFDA Q8, Q9, Q16, Q25
FDA questions it names
Questions this filing names by number.
Q24 · Third-party foundation model changes
Across the five cross-cutting questions
RecovryAI’s reading of the whole filing. Silence is never counted as opposition.
Patient-facing autonomyShould FDA permit patient-facing AI to act with meaningful autonomy within a defined scope?
No position stated
Proportionate evidenceShould evidence requirements scale with clinical risk rather than a uniform high bar?
No position stated
Postmarket relianceCan strong postmarket monitoring justify accepting more premarket uncertainty?
No position stated
Competency evaluationCan a device be evaluated on competency benchmarks and clinical confirmation against clinicians?
No position stated
Change controlCan devices on third-party foundation models be maintained under pre-specified change control?
Supports with conditions
Autonomy acceptedThe highest level this filing accepts
Low-consequence work: Not stated
High-consequence work: Not stated
This filing was submitted by a member of the public. We publish our analysis here and link to the submission rather than reproducing it, and this page is not indexed by search engines.
Read the full submission on regulations.gov ↗