The principle that organisations and individuals should be responsible and answerable for the impacts of the AI systems they develop, deploy, or operate. Accountability encompasses both preventing harms through responsible design and providing remedies when harms occur.

AI accountability establishes clear responsibility for AI outcomes, ensuring that those who develop and deploy AI systems take appropriate care in their design and use. Effective accountability requires clear lines of responsibility within organisations, documented decision-making processes, impact assessment procedures, and mechanisms for addressing issues when they arise. Technical measures like audit trails and monitoring systems support accountability by providing evidence of system behaviour and human oversight.

‍

A financial institution establishing accountability for an automated credit scoring system by designating specific roles responsible for fair lending compliance, implementing regular bias audits, maintaining detailed documentation of design decisions, and providing clear appeals processes for customers who believe they’ve been incorrectly assessed.