The formal right held by a governance body or stakeholder to block or require changes to AI deployments that pose unacceptable risks.
A governance mechanism granting specific roles (e.g., AI Ethics Board chair, Chief Risk Officer) the power to halt or demand modification of AI projects that fail to meet risk or ethics thresholds. Veto decisions are recorded, justified with documented rationale, and trigger action plans to address identified issues before any resubmission.
A retail bank’s AI Steering Committee reserves veto authority. When a pilot of a customer-scoring model exhibits demographic parity below the acceptable threshold, the committee exercises its veto, pausing deployment until bias remediation measures are implemented and revalidated - ensuring no high-risk model goes live without committee approval.
