Bias Audit of Morgan Stanley’s Eightfold Model

This report details the findings of a bias audit conducted by BLDS, LLC, on Morgan Stanley's use of the Eightfold algorithmic model for scoring job applicants. The audit was performed in compliance with New York City’s Automated Employment Decision Tools ordinance. Morgan Stanley provided applicant records for the period from June 1, 2024, to May 31, 2025, encompassing 956,931 applications. The audit computed impact ratios for various demographic categories as mandated by the ordinance. The results indicated that the Eightfold model does not produce a disparate impact on any protected group, with impact ratios ranging from 1 to 0.869, all above the threshold of 0.80 set by federal guidelines. The report also suggests that a more precise bias audit could be conducted at the job level, which was also analyzed, yielding similar results. The findings are intended to ensure compliance with legal standards regarding automated employment decision tools.