# Liability Laundering **Domain:** AI Governance / Accountability / Institutional Design **Doc Type:** Established Governance Critique **Maturity:** Developed ## Definition **Liability laundering** is the transfer of apparent responsibility from the institutions that design, procure and deploy an automated system to the downstream person who formally approves its output. The organization receives the efficiency and authority of automation while preserving a human signature to absorb blame when the system fails. ## Structural Pattern - upstream actors choose data, objectives and thresholds; - the interface presents a recommendation as authoritative; - workload and [[wiki/Automation Bias|Automation Bias]] reduce meaningful review; - the operator retains nominal discretion but lacks information or power to reconstruct the decision; and - accountability is assigned to the operator after harm occurs. ## Constitutional Response Responsibility must follow control over the model, data, deployment conditions and appeals process. [[wiki/Fiduciary Governance|Fiduciary Governance]] and [[wiki/Algorithmic Constitutionalism|Algorithmic Constitutionalism]] require accountable owners for each layer rather than treating human presence as a complete governance strategy. ## Key Insight **Human presence cannot legitimate an automated decision when the human lacks the conditions required to judge it.**