The Uniform Guidelines on Employee Selection Procedures have governed federal hiring since 1978. Their core demand was modest and mathematical: if an agency uses a test, a screen, or a selection procedure that filters out one group at a higher rate than another, the agency must analyze that adverse impact and validate that the procedure actually predicts job performance. Show your work, in other words, before the work shows bias.
As of today, that demand is gone from federal personnel regulation. An interim final rule from the Office of Personnel Management, published this morning and effective on publication, removes every reference to the guidelines from 5 CFR parts 300, 330, 337 and 720, the provisions covering employment practices, career transition, examination and equal opportunity. [1]
There was no advance comment period. OPM invoked 5 U.S.C. 553(b)(B), the exception that lets an agency skip notice and comment when it finds them "impracticable and contrary to the public interest." [1] The finding here is that the guidelines' legal foundation was voided by a June 9, 2026 opinion of the DOJ Office of Legal Counsel, which concluded that existing interpretations of Title VII "embrace an unconstitutional reading of Title VII insofar as they contemplate liability based on disparate effects alone," an opinion made binding on every agency by Executive Order 14215. [1]
What survives, and it matters: agencies must still perform job analysis, selection procedures must still bear a rational relationship to job performance, general nondiscrimination provisions remain, and Title VII itself, being a statute, is untouched by any regulation's repeal. [1] Intentional discrimination remains illegal everywhere. The steel-man for the rule is genuine: if disparate-effects liability is unconstitutional, as the OLC opinion holds, then regulations built on it lack a legal floor, and an agency cleaning its books is doing administrative hygiene, not policy.
The record beside that argument is what the deleted machinery actually did. Validation studies and adverse-impact analyses were not liability findings. They were measurements, the numbers that told an agency, and any court or applicant later, whether a screen was filtering people by ability or by demography. Removing the mandate does not resolve the constitutional question the OLC opinion raises. It removes the data that would let anyone litigate it. An applicant screened out by an unvalidated test now faces a record in which no one was required to notice.
Yesterday we reported that the Education Department's retreat from disparate impact was in fact nine agencies wide, executed through the Federal Register while coverage looked elsewhere [callback]. Today the government's own hiring rules joined the sweep, effective before the ink dried. Two days, one direction. The comment period, such as it is, now happens after the fact, and we will report what lands in it.
