The United States Patent and Trademark Office (USPTO) has granted patent US12585956B2 to SAS INSTITUTE INC. for a novel system, method, and computer-program product designed to mitigate bias in machine learning-based decisioning models. The patented technology involves obtaining decisioning datasets, using machine learning algorithms to detect bias by comparing favorable and unfavorable decisioning records, executing counterfactual assessments, and generating explainability artifacts that detail the bias intensity. These artifacts are then displayed in a user interface.

This patent grant signifies a technological advancement in addressing algorithmic bias, particularly relevant for organizations developing or deploying AI-driven decisioning systems. While patents do not impose direct regulatory obligations, they represent proprietary technology that could influence future industry standards or product development. Companies utilizing machine learning for critical decisions, such as loan applications, hiring, or risk assessments, should be aware of this patented approach, as it may impact the competitive landscape and the availability of certain bias mitigation techniques.