Human capital management (HCM) systems are among the highest stakes domains for the deploy‐ ment of artificial intelligence in enterprise software. The decisions HCM systems influence (selec‐ tion, compensation, performance evaluation, leave administration, benefits adjudication) directly af‐ fect employee livelihoods and are subject to a growing body of regulatory attention. Despite the stakes, governance frameworks targeted to the operational realities of HCM AI deployment remain underdeveloped. This paper offers a risk based governance framework specifically calibrated for HCM systems incorporating large language models, retrieval augmented generation, and other gen‐ erative AI components. The framework defines four risk classes, six governance domains, and a set of operational controls intended to be enforceable by engineering, product, and compliance teams without requiring abstract principles to be re translated for each new use case. The framework draws on direct experience leading the deployment of multiple production AI systems in a large enterprise HCM platform, supplemented by alignment with the NIST AI Risk Management Framework, the European Union Artificial Intelligence Act, and emerging United States state level employment arti‐ ficial intelligence regulation. The framework is intended for HCM software engineering and product leaders, in house compliance professionals, and HR technology executives.
Prashanth Reddy Pasham (Mon,) studied this question.