A public-health board requires new GenAI features to pass an organizational Responsible AI checklist before go-live. Which governance action best implements that requirement?

Aligning each GenAI use case with the board’s Responsible AI policy checklist before production release is the governance action that implements a public-health board’s requirement that new features pass an organizational Responsible AI checklist before go-live. The checklist is a release gate: map the feature’s purpose, data classes, affected populations, fairness/privacy/transparency controls, and human oversight to explicit policy items, record evidence, and obtain approval before traffic reaches clinics or residents. That sequencing keeps Responsible AI from becoming post-hoc paperwork.

Skipping the checklist for pilots that “feel obviously helpful” to clinic staff creates shadow production risk—pilots often touch real patients and data. Treating Responsible AI policy as optional marketing copy after launch inverts the control: policy must constrain design and release, not decorate a landing page. Deploying first and scheduling the checklist workshop only if a complaint arrives is reactive governance that fails the board’s before-go-live requirement.

Exam caveat: organizational RAI frameworks are pre-production gates, not optional post-launch content. Exam distractors often romanticize “helpful pilots” or complaint-driven review; the correct pattern is checklist alignment before release.

Operational check: bind the public-health GenAI CI/CD or change board to a completed RAI checklist artifact, refuse production promotions without it, archive signed checklist versions with the feature ID, and re-run the checklist when data sources, populations, or model providers change.