FDA Seeks Input on “Non-Device” Clinical Decision Support and AI
FDA is gathering public comments for a 2026 report on patient-safety best practices related to certain “non-device” software functions under the 21st Century Cures Act—an issue that matters when AI-enabled tools provide clinical recommendations. Comments are due August 13, 2026.
FDA is collecting public input for a 2026 report on patient-safety impacts of certain “non-device” software functions under the 21st Century Cures Act—including some software intended to provide limited clinical decision support. FDA updated this solicitation on July 14, 2026, and submissions are due August 13, 2026 under docket FDA-2018-N-1910.
For patients, caregivers, and clinicians-adjacent readers, the key takeaway is that FDA oversight depends on how software is intended to function in care—not simply on whether it uses AI or similar technologies.
What FDA is asking the public to weigh in on
In its update, FDA describes categories of “non-device software functions” it plans to consider as part of its report, including software intended for things like administrative support, encouraging a healthy lifestyle, electronic patient records, and software that transfers, stores, converts, or displays data—as well as software described as providing limited clinical decision support.
FDA is asking for patient-safety best practices and related input as part of how these functions may be used in real-world care settings.
How FDA distinguishes “non-device” clinical decision support
FDA’s Clinical Decision Support materials explain how the agency interprets the Cures Act pathway for “clinical decision support software functions” that can be excluded from the medical device definition, using criteria meant to focus on what the software is doing and how it supports (or doesn’t replace) clinical judgment.
Examples of common safety boundary themes
- Simple calculations: FDA explains that software intended to perform certain simple medical calculations commonly used in clinical practice generally does not fit the “non-device” clinical decision support framework.
- Clinical recommendations: FDA also discusses how software that presents clinical recommendations is evaluated based on the overall intended function and how it fits within the non-device criteria.
- Time-critical decision making: FDA states that software intended to support time-critical decision making does not meet the definition of non-device clinical decision support. Separately, FDA explains that the context and the software’s primary function (for example, helping display relevant patient information at the point of care) can matter.
Bottom line: a tool’s role in the workflow—whether it mainly displays information versus makes (or meaningfully steers) clinical decisions—can affect where it lands in FDA’s framework.
Why this matters for AI-enabled tools
Even when software is not regulated as a device, patients and families deserve clarity about how information is produced and how decisions are reviewed.
The CDC’s guidance on disclosing generative AI use (for example, in scientific contexts) emphasizes transparency and human accountability. In a separate policy statement, the American Medical Association similarly emphasizes that AI should support—not replace physician judgment, and that accountability and clinician oversight are central when AI is used in care.
What readers can do now (practical questions)
- Ask the care team: Is any AI-enabled software providing recommendations, risk flags, or decision support? Who reviews or verifies what the tool suggests?
- Clarify the tool’s role: Does it primarily display patient information, or does it push clinical actions (especially in time-sensitive situations)?
- Request plain-language explanation: What data sources and logic are used, and what happens if a clinician disagrees with the tool’s output?
- For clinician-adjacent readers: consider whether the tool’s intended function aligns with the “non-device” criteria FDA discusses—particularly around time-critical use and how recommendations are presented.
What’s known vs. not yet known: FDA’s current step is collecting input to inform its upcoming 2026 report. That does not automatically mean any specific AI product will change immediately. The report effort is meant to improve understanding of patient-safety best practices for these software functions.
What’s next
FDA will use submitted perspectives to develop its report focused on risks and benefits related to non-device software functions, including limited clinical decision support. If you’re affected by these tools in care, the most reasonable next step is to keep questions focused on what the software is doing, how clinicians oversee it, and how final decisions are made.
Sources
- FDA — Reports on Non-Device Software Functions (Update: July 14, 2026)
- CDC — Considerations for Disclosing Generative AI Use in Scientific Work (May 28, 2026)
- American Medical Association — Policies to ensure AI supports—not replaces—physician judgment (Jun 10, 2026)
Editorial note: Weence articles are researched from cited public-health, medical, regulatory, journal, and reputable news sources and may be drafted with AI assistance. They are checked for source support, clarity, and safety guardrails before publication.
This article is for general informational purposes only and is not medical advice. Research findings can be early or incomplete, and health guidance can change. Always talk with a qualified healthcare professional about personal symptoms, diagnosis, medications, vaccines, screenings, or treatment decisions. If you think you may have a medical emergency, call emergency services right away.
