Be compliant for California’s FEHA rules

The amended FEHA regulations apply to employers with five or more employees that use automated-decision systems (ADS) in California employment decisions — hiring, promotion, training selection, and more. They also reach the vendors and “agents” acting on an employer's behalf, so both the buyers and builders of employment AI carry responsibility. The rules took effect October 1, 2025.

An ADS is any computational process that makes or assists an employment decision. In practice that includes resume screeners and automated filtering tools, AI-powered video interview scoring, hiring or ranking recommendations, and performance, promotion, or termination tools. If a system meaningfully influences who gets hired, advanced, or let go, FEHA likely treats it as an ADS.

No — FEHA does not mandate a bias audit. But the regulations make the quality, scope, and recency of bias testing central to liability: proactive anti-bias testing can be raised as a defense to a discrimination claim, while the absence of testing can be used as evidence against the employer. Independent bias audits from Warden AI give employers and vendors that defensible, documented record.

At least four years. That covers the data fed into the tool, the outputs it generated such as scores or rankings, the criteria applied to candidates, and the results of any bias testing or evaluation. The amendments also extended general personnel-record retention to four years, up from two.

Yes. The regulations expand the definition of “employer” to include agents and third parties that act on an employer's behalf — including those that develop or deploy an ADS. That means a vendor supplying an employment AI tool can be held jointly liable alongside its client for discriminatory outcomes.

FEHA prohibits employment discrimination across 18 protected classes, including race, color, national origin, ancestry, religion, sex, gender identity and expression, age, disability, medical condition, genetic information, marital status, sexual orientation, and military or veteran status. A meaningful bias evaluation should test outcomes across these groups.

Employers are expected to tell candidates and employees when an ADS is used to make or influence an employment decision, and to provide reasonable accommodations on request. Clear notice, combined with human oversight of high-stakes decisions, is part of the “reasonable safeguards” the rules expect.

Warden AI provides independent bias audits, time-stamped audit trails, and dashboards and reporting aligned to the regulations — documenting the proactive testing and safeguards FEHA expects. Whether you are a vendor testing pre-market or an employer monitoring a live system, Warden AI gives you evidence of fairness you can show regulators, courts, and procurement teams.