Complying with California’s New AI Hiring Laws: CRD Regulations Explained

1. Background & Regulatory Timeline

California's Civil Rights Council, a branch of the California Civil Rights Department (CRD proposed regulations in May 2024 to address discrimination risks related to automated-decision systems (ADS) used in employment [1]. On March 21, 2025, the final version of the “Employment Regulations Regarding Automated-Decision Systems” was approved by the Civil Rights Council. These regulations are expected to take effect on October 1, 2025, following formal approval by the Office of Administrative Law [2].

2. What Qualifies as ADS?

The CRD defines automated-decision systems broadly. These include the use of tools such as algorithms, artificial intelligence, or machine-learning to make or assist in employment decisions, including resume screening, facial analysis, voice profiling, and predictive hiring models. [2]. Employers are responsible for these systems whether they are developed internally or sourced from third-party vendors, who are also directly accountable under the rules [1][2].

3. Prohibited Use & Bias Concerns

Disparate Impact and Proxy Discrimination
The CRD prohibits employers from using ADS in ways that disproportionately affect protected groups, including those based on race, gender, age, disability, language proficiency, height, and weight or other protected characteristics [2][3]. Criteria that seem neutral but correlate with protected characteristics—like tone of voice or educational background—may also constitute illegal discrimination [2].

Restricted Evaluations
Tools that evaluate an individual’s health, psychological characteristics, or criminal history are highly regulated and may only be used when they meet strict standards related to job relevance and nondiscrimination [2][3].

4. Compliance Requirements

Recordkeeping
Employers and vendors must retain documentation related to ADS and must be kept separately from personal flies, including design notes, datasets, applicant outcomes, and bias testing reports—for at least four years from the date of making the record or the date of the personnel action occurred [2].

Anti-Bias Testing
Anti-bias testing includes quality, efficacy, recency, and scope of effort, results of testing and response to results. Employers are required to conduct and document statistical analyses to identify disparate outcomes. These evaluations are essential to defending against any potential claims under California’s Fair Employment and Housing Act (FEHA) [2][3].

Human Oversight
All ADS-based decisions must include meaningful human review. Employers remain fully responsible for employment decisions and cannot shift legal liability to an algorithm or system. The presence of AI does not diminish or replace an employer’s obligations under civil rights laws. [2].

Transparency & Appeals
Proposed legislation such as Senate Bill 7 may require employers to notify applicants when ADS is used, employers should clearly inform applicants when ADS will be used and how they can request an accommodation and provide them with a meaningful opportunity to contest automated decisions [5].

5. Liability & Enforcement

Shared Liability
Both employers and third-party vendors can be held liable if the use of ADS results in discrimination. These regulations emphasize joint accountability for the impact of hiring technologies [2][3].

Enforcement Tools
The CRD may investigate potential violations and impose penalties under FEHA, including injunctive relief, back pay, reinstatement, and monetary damages. The agency is also expanding its capacity to audit ADS practices across industries [3].

6. Real-World Bias Risks

A June 2025 study revealed that leading AI hiring systems—including those built on large language models—exhibited measurable bias, with significantly different recommendation rates based on race and gender, even when résumés were equivalent [4]. These findings highlight how subtle design choices or biased datasets can lead to unintended discrimination, despite the absence of overtly discriminatory criteria.

7. Best Practices for Employers

1. Conduct an ADS inventory to identify what systems are used and where.

2. Run regular adverse impact analyses using reliable statistical methods.

3. Maintain thorough records of all data inputs, outcomes, and decisions.

4. Ensure human review is embedded in each step of the hiring process.

5. Hold vendors accountable by requiring compliance terms in contracts and requesting transparency into their tools.

   
   
   

Final Thoughts

California’s CRD regulations signal a major shift in how employers must manage Artificial Intelligence -powered hiring practices. By focusing on transparency, fairness, and accountability, the rules aim to balance technological innovation with civil rights protections. As artificial intelligence becomes more embedded in the workplace, ensuring compliance with these standards is critical for both legal protection and ethical hiring.

Sign Up for Our Upcoming Webinar Mid-Year Compliance in 2025: Critical Updates for Businesses

This article is not intended to be exhaustive nor should any discussion or opinions be construed as legal advice. Readers should contact legal counsel for legal advice.

Works Cited

1. California Civil Rights Council. (2024). Notice of Proposed Rulemaking – Automated-Decision Systems. https://calcivilrights.ca.gov
2. California Civil Rights Department. (2025). Final Text of Proposed Regulations on Automated-Decision Systems. https://calcivilrights.ca.gov
3. California Fair Employment and Housing Act (FEHA). https://leginfo.legislature.ca.gov/faces/codes.xhtml
4. New York Post. (2025, June 24). AI-powered hiring tools favor Black and female job candidates over white and male applicants: study. https://nypost.com/2025/06/24/business/ai-hiring-tools-favor-black-female-candidates-over-whites-males
5. California State Senate. (2025). Senate Bill 7 – Automated Decision Notification Requirements. https://leginfo.legislature.ca.gov