As I catch up on reading in clinical research, regulatory science, and bioethics, two publications caught my attention because they connect with questions I have examined through my own analysis of AI, informed consent, patient-data governance, and the regulatory requirements governing clinical research.

Adebamowo et al. (2026) examine why documented permissions and data provenance alone may not establish continuing ethical justification for secondary uses. Hovhannisyan and Boyd (2026) question one-time consent as AI evolves in clinical care.

As AI expands what we can learn from clinical data, are we giving equal consideration to what patients understand about its potential uses? Information collected for care or research may later support new studies, AI model development, or applications not anticipated at collection. The issue is not simply whether data can be reused, but whether the proposed use remains appropriately justified and governed.

Not every secondary use requires renewed consent; requirements depend on the data, purpose, permissions, and legal framework. An ethical concern does not automatically establish a regulatory violation. Legal permissibility and meaningful patient understanding are related, but distinct.

In my view, responsible governance calls for a documented, risk-based assessment: Is the new purpose consistent with the original collection context and applicable permissions? Have data transformations or linkages altered privacy risks or the meaning of the information? Do material changes warrant additional ethics, data-access, or organizational review—and who is accountable?

This need not mean reopening every routine use. Oversight should be proportionate to meaningful changes in purpose, risk, and downstream application, with safeguards tailored to the concern identified. AI can advance research, but human judgment and accountability must remain central. Progress is measured not only by what technology makes possible, but by how responsibly we use it.

How should clinical research organizations decide when a new AI-enabled use of existing health data warrants additional scrutiny? References –Adebamowo, C., Adebamowo, S. N., Akintola, A., Ikhane, P., Akintola, S., Ogundiran, T., Jegede, A., Adeyemo, O., Callier, S., Imam-Tamim, M., Uthman, I., & BridgELSI Project as part of the DS-I Africa Consortium. (2026). The continuity trap in data science health research. Journal of Medical Internet Research, 28, e98699. https://lnkd.in/e-Drfs5W –Brothers, R. (2026). Ted talk: My thoughts on the challenges with the utilization of AI in clinical research [Unpublished manuscript]. Arizona State University. –Hovhannisyan, M., & Boyd, J. W. (2026). Informed consent in the age of clinical AI: Beyond the one-time authorization model. The American Journal of Bioethics, 26(9), 9–13. https://lnkd.in/evcjH4Wm