Integrating opaque technologies into business processes complicates the governance of ethical risks. This paper addresses the Ethical Risk Handover, defined as the structural loss of normative intent during the transition from design to technical configuration. We argue that this discontinuity transforms abstract values into concrete operational threats: privacy oversights harden into compliance risks, unchecked algorithmic bias manifests as legal liability, and opaque decision-making creates reputational threats. To resolve this, we introduce an automated framework that leverages Large Language Models to detect latent risks. We operationalize these findings via a Camunda Modeler plugin that projects risk scores as a heatmap overlay directly on the process model. Evaluation against real-world processes confirms that the system effectively interprets implicit risks, achieving detection rates comparable to domain experts and restoring the visibility required to mitigate operational threats.