Quis Custodiet Ipsos Custodes? — From Asimov to ChatGPT and Beyond

Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).

Accepted author manuscript — GrandRoundsMD accepted this essay for publication on September 25, 2026. Publication by GrandRoundsMD has not been independently verified as of October 4, 2026. This public Zenodo deposit preserves the unchanged September 24, 2026 submission copy; it is not a separately counted article or a GrandRoundsMD version of record. Peer-review status has not been established. Using Asimov’s Three Laws and Zeroth Law as an organizing analogy, this essay asks what human control means when AI supplies both a recommendation and the information used to judge it. It argues that meaningful control requires independently accessible evidence and an effective power to stop AI actions, with enforceable limits and accountable institutions. The PDF and Word files are the preserved submission copies. AI tools assisted with editorial review, source checking, and formatting; the author reviewed and approved the submission.