Risk-Tiered Governance of Generative AI in Anatomy Education: Donor Material, Fidelity, and Assessment

A proposed governance framework for GenAI in anatomy education uses four risk tiers (Low, Moderate, High, Restricted/Avoid) under a conservative rule—the highest triggering criterion sets the tier. Permitted uses then pass five control gates: inputs/provenance, anatomical fideli…

• Tier assignment follows a conservative rule: purpose, audience, input/data class, consequence, scale, and AI autonomy each trigger a tier, and the highest among them governs the use case.
• Five control gates must be cleared: inputs/provenance, anatomical fidelity, variation/representation, assessment validity (where relevant), and disclosure/accountability.
• Four core decisions are never delegated to AI; approval is not a one-time authorization but a revisable state maintained through monitoring, incident-response, and revalidation.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.