Temporal CEO on AI safety concerns: We should 'absolutely' audit what the agents have been doing