Users should have accessible mechanisms to identify and report harmful or concerning AI system incidents and impacts, with such warnings shareable among relevant stakeholders. Feedback should be continuously incorporated into system updates and communicated to relevant stakeholders.