A utilities control DC must limit which datasets may leave the municipal boundary during hybrid training. Which hybrid evaluation is most important?

A utilities control DC that must limit which datasets may leave the municipal boundary during hybrid training should enforce data-governance rules that define which datasets may leave for hybrid training. Governance is the control plane for hybrid egress: classified allow-lists, blocked regulated telemetry, and exceptions that are explicit—not set by defining which models may return from hybrid training, which controls inbound artifacts rather than which datasets may leave.

Defining which models may return from hybrid training is a return-ingress control, not a data-governance rule for which datasets may leave. Defining how long datasets may persist for hybrid training is a retention-lifecycle rule, not a boundary-control decision about which datasets may leave. Defining which fields may be masked for hybrid training is a field-level de-identification rule, not a data-governance rule for which datasets may leave.

Exam caveat: Domain 2.5 includes evaluating data synchronization and integration under policy; hybrid training eligibility is an egress-allowlist decision first, return, retention, or masking decision second.

Operational check: maintain an authoritative egress catalog for utilities datasets, block non-approved hybrid training submissions at the orchestrator or DLP boundary, and audit cloud landing zones for unexpected municipal telemetry. Utilities operational technology adjacent datasets are especially sensitive to casual hybrid egress; governance must distinguish derived features cleared for cloud training from raw SCADA or customer-linked telemetry that must remain inside the municipal boundary regardless of which models are approved to return from hybrid training.