Speakers described "guardrails" as practical controls that define what AI systems can do, how they access data and when human oversight is required. In the discussion, trust was closely linked to ...