California FEHC AI Employment Regulations 2026: ADS Compliance Guide
California employers using AI to screen resumes, rank candidates, or flag employees for discipline now face a Civil Rights Council rulebook built specifically for automated-decision systems. It doesn't replace federal exposure — it stacks on top of it. Here's what changed and what to do about it.
What the Civil Rights Council Actually Did
California's Civil Rights Council — the rulemaking body for the Fair Employment and Housing Act (FEHA) — adopted regulations that spell out, in unusually specific terms, how existing employment-discrimination law applies when an automated-decision system (ADS) is involved in a hiring, promotion, discipline, or termination decision. The core message to employers: plugging an algorithm into a decision doesn't change who's liable if that decision discriminates.
This isn't a brand-new statute creating new causes of action. It's a clarification of how disparate-treatment and disparate-impact theories — the same ones that have always governed hiring decisions — apply when the "decision-maker" is a resume screener, a video-interview scoring tool, or a scheduling algorithm.
Who and What Is Covered
- Any automated tool used in an employment decision — resume screening, candidate ranking, video-interview analysis, scheduling, performance scoring, and discipline/termination recommendations.
- Employers of covered size under FEHA — generally businesses with five or more employees, the same threshold that applies to FEHA generally.
- Third parties acting as agents — staffing agencies, recruiting platforms, and software vendors that perform employment functions on an employer's behalf can be pulled into the same liability analysis.
- Every stage of the employment lifecycle — not just hiring. Promotion, compensation, discipline, and termination decisions made or assisted by automated tools are covered.
Recordkeeping: The Part Employers Underestimate
The regulations extend recordkeeping expectations to cover the inputs, criteria, and outputs of automated-decision systems — not just the final hiring decision. In practice, that means employers need to preserve enough data about how an ADS tool scored or ranked candidates and employees to reconstruct and defend the decision later, for a retention period well beyond most companies' informal habits. Employers who can't produce this data when a charge is filed are in a materially worse position than those with a documented audit trail.
Vendor Liability: You Can't Fully Outsource the Risk
Following the same pattern seen in federal litigation over AI hiring tools, California's rules contemplate that a vendor or third party performing employment functions can be treated as an agent of the employer — extending potential liability to the software provider itself in some circumstances. For employers, this means vendor contracts, indemnification clauses, and validation documentation from ADS providers are now a genuine risk-management tool, not boilerplate.
How This Stacks With Federal and Other State Law
California employers don't get to choose between federal and state exposure — both apply simultaneously, plus any local rules:
- Title VII, the ADEA, and the ADA continue to apply at the federal level regardless of state rules.
- EEOC guidance on the four-fifths rule remains a relevant screening heuristic for adverse impact.
- NYC Local Law 144 and Illinois' video-interview law apply independently to employers hiring in those jurisdictions.
- Colorado's AI Act imposes its own high-risk employment obligations for employers hiring there.
Compliance Checklist for California Employers
Treat every ADS touchpoint in the employee lifecycle as a decision that must be defensible.
Don't let compliance gaps stack up across your web presence
Candidate-facing career sites and application portals carry accessibility risk on top of employment-law risk. RatedWithAI scans your site for the issues that turn into complaints.
Scan Your Site for Free →Frequently Asked Questions
Does this create a brand-new lawsuit right for California employees?
No. It clarifies how the existing FEHA framework — disparate treatment and disparate impact — applies when an automated-decision system is involved. The underlying causes of action aren't new; the application to algorithmic tools is.
How is this different from New York's Local Law 144?
NYC LL144 imposes a specific, standalone requirement: an independent bias audit published before use of an automated employment decision tool. California's rules work through the broader FEHA disparate-impact framework and extended recordkeeping instead of a bright-line audit-and-publish mandate — but the practical effect (needing defensible bias evidence) is similar.
Do these rules apply to small California employers?
FEHA generally applies to employers with five or more employees, which is the threshold that also governs these ADS-specific clarifications. Very small employers below that threshold fall outside FEHA's scope, though federal law may still apply.
What's the single biggest compliance gap employers have right now?
Recordkeeping. Most companies don't retain the ADS inputs, scoring criteria, or output data needed to reconstruct and defend a decision months or years later. Fixing that — even before running a full bias audit — closes the most common gap.