Key Notes:
- California’s Civil Rights Department approved regulations on the use of artificial intelligence and automated-decision systems in employment.
- The new rules clarify how existing antidiscrimination laws apply to AI tools.
- Employers must retain employment and automated-decision data for at least four years.
- Certain AI-based assessments may constitute unlawful medical inquiries.
- Regulations took effect on October 1, 2025.
The California Civil Rights Department (CRD) is charged with enforcing many of the state’s robust civil rights laws, including those in the areas of employment, housing, businesses and public accommodations, and state-funded programs and activities. On June 30, 2025, the California Civil Rights Council, which issues regulations, announced final approval of regulations that address the use of artificial intelligence (AI), algorithms, and other automated-decision systems in employment.
Why Now?
Automated-decision systems are increasingly used for making a wide range of employment decisions. Employers assume that these tools take certain decisions, like those related to recruitment, hiring, and promotion, out of their hands and make the company one step removed from discrimination claims by applicants. While these tools can streamline processes, the CRD notes that their use can still replicate or exacerbate existing biases, for example, by mimicking a male-dominated workforce or steering ads based on gender or race, leading to discriminatory outcomes. Also, an AI-powered hiring assessment platform may use online tests, games, or questionnaires to evaluate job applicants’ cognitive abilities, reaction times, or physical skills. If such systems are designed without considering accessibility or reasonable accommodations, they may inadvertently screen out individuals with disabilities.
What Do the Regulations Do?
The new regulations clarify how California’s existing antidiscrimination laws apply to AI and automated-decision systems in the workplace. Among other provisions, they:
- Confirm that use of an automated-decision system can violate the law if it harms applicants or employees based on protected characteristics (e.g., gender, race, disability).
- Require employers and covered entities to maintain employment records – including automated-decision data – for a minimum of four years.
- Affirm that automated assessments (e.g., tests, questions, puzzles) that elicit information about a disability may constitute an unlawful medical inquiry.
- Add definitions for key terms, including “automated-decision system,” “agent,” and “proxy.”
What This Means for Employers and Vendors
Employers must ensure that the criteria they input into an algorithmic or AI-based tool is nondiscriminatory. This includes making sure all questions, puzzles, or assessments avoid eliciting disability-related information and take into account reasonable accommodations for those who may be disabled. Employers also must now ensure internal policies align with the regulations’ defined terms (“automated-decision system,” “agent,” “proxy”).
The use of AI tools in employment decisions has been a growing focus for regulators, advocacy groups, and legal commentators, and several investigations and complaints have referenced algorithmic bias and automated assessments. We expect an increase in formal complaints and litigation involving AI-based employment decisions, especially as awareness grows and recordkeeping requirements make it easier to investigate and prove discriminatory impact.
For businesses and employers, this development underscores the need to closely monitor state-level regulatory trends and prepare for compliance with differing requirements across jurisdictions. While comprehensive federal AI legislation remains under development, companies must be prepared to navigate varying state laws governing the use of AI in employment decisions. Consistent with this new California law, federal EEOC guidance provides that when an employer utilizes AI to make an employment decision and the AI use results in an employment decision based upon a legally protected characteristic, the employer can be held liable for discrimination in violation of federal antidiscrimination laws. Under the new California legislation, if an employer’s AI use results in an employment decision that legally discriminates against someone based upon a legally protected characteristic, they also may be liable for discrimination under California law.
Action Steps
Employers should proactively review their AI tools and practices to mitigate risk and ensure compliance.
- Take inventory: Identify all AI/automated-decision tools used across the employment lifecycle (including vendor tools).
- Assess impact: Evaluate for potential bias or disparate impact on protected groups; document testing and results.
- Update policies: Incorporate AI governance into EEO, hiring, promotion, accommodation, and privacy policies.
- Evaluate vendor management: Amend contracts to require transparency, testing, data retention, cooperation with audits, and allocation of risk and liability in the event that the use of an AI vendor’s software results in illegal discrimination.
- Update data retention policies: Implement systems to retain relevant employment and automated-decision data for four years.
- Provide training: Train HR, recruiters, and managers on appropriate use, limitations, and legal risks of AI tools.
- Consider accessibility and accommodation: Review assessments for potential unlawful medical inquiries; provide reasonable alternatives as required by law.
This advisory bulletin may be reproduced, in whole or in part, with the prior permission of Thompson Hine LLP and acknowledgment of its source and copyright. This publication is intended to inform clients about legal matters of current interest. It is not intended as legal advice. Readers should not act upon the information contained in it without professional counsel.
This document may be considered attorney advertising in some jurisdictions.
© 2025 THOMPSON HINE LLP. ALL RIGHTS RESERVED.
