Auditing Algorithms: What New York’s Bias Audit Mandate Reveals about the Future of Artificial Intelligence Assurance

SYNOPSIS

AI systems are increasingly used to support consequential organizational decisions, heightening concerns about fairness, reliability, and accountability. Regulators have begun to respond through audit and disclosure mandates, yet credible assurance over AI systems remains underdeveloped and it is unclear who should provide it. This commentary argues that algorithmic assurance is a natural extension of the assurance function and that accounting professionals are well positioned to deliver it because they bring independence, evidence discipline, professional judgment, and structured reporting. Using New York City’s Local Law 144 as an illustrative case, this study analyzes 13 publicly available bias-audit reports. The analysis shows that audit concepts such as scope, criteria, evidence, and independence are already being applied to algorithms, largely outside the accounting profession, but with wide variation in rigor and transparency. The commentary identifies the core design problems behind this variation and discusses implications for audit firms, regulators, and standard setters.

Data Availability: Data are available from the public sources cited in the text.

JEL Classifications: M42; M48; O33.