Effective October 1, 2025, California will enforce new anti-discrimination regulations targeting the use of AI and automated technologies in employment decisions. This marks a significant shift in how businesses across the state must approach recruitment, screening, interviewing, and employee evaluation processes—especially if they use AI or software-assisted tools.
What’s Changing: AI in Employment Now Under Scrutiny
The California Civil Rights Council has broadened its anti-discrimination rules under the Fair Employment and Housing Act (FEHA) to include “Automated-Decision Systems”—even when humans remain involved in final decision-making.
Definition of “Automated-Decision System”
An Automated-Decision System (ADS) is any computational tool—using AI, algorithms, machine learning, or statistical models—that either makes decisions or assists humans in making decisions related to employment benefits, such as hiring, promotions, or layoffs.
Key tasks now regulated under this definition include:
- Screening and ranking applicants
- Directing job ads to targeted groups
- Resume filtering for keywords or qualifications
- Analyzing facial expressions, tone of voice, or word choice in online interviews
- Using third-party data to evaluate candidates
Note: Even tools that only assist in decision-making—not just fully autonomous systems—are covered under this regulation.
Employer Prohibitions and Legal Risks
Employers are prohibited from using automated tools or selection criteria that result in discrimination based on characteristics protected under the FEHA, including race, gender, national origin, disability, and more.
Specifically, businesses may not use:
- Pre-employment inquiries that indirectly identify protected characteristics
- AI tools that inadvertently discriminate during interviews (e.g., voice tone or facial analysis)
- Screening systems that fail to accommodate individuals with disabilities
- Biased algorithms or datasets that filter candidates unfairly
Defense Strategies
If an employer faces a discrimination claim related to AI usage, evidence of anti-bias testing, controls, and oversight can serve as a defense. Employers must be able to show that they:
- Regularly test AI systems for discriminatory outcomes
- Address and document results from those tests
- Implement quality assurance and ethical AI practices
Expanded Record Retention Requirements
One of the most impactful changes is the doubling of record retention requirements—from two years to four years—for any employment practice involving automated systems.
What Must Be Retained:
- Applications and resumes
- Interview records and evaluations
- Selection criteria
- Data used by, in, or to customize an Automated-Decision System
- All data generated during or after use of an AI tool in recruitment or evaluation
For high-volume employers, this creates significant operational and compliance implications. Screening thousands of candidates means retaining massive volumes of personal and decision-related data—often sensitive and subject to additional data privacy requirements.
Tip: Employers should update contracts with vendors to include data-sharing, storage, and retention clauses that align with these new rules.
How This Impacts Employers
HR, legal, and IT teams must act quickly to ensure AI tools and recruitment software meet California’s October deadline. Key actions include:
- Audit all systems used in recruiting or employee evaluation for potential AI involvement
- Evaluate vendors and software platforms for compliance and recordkeeping capabilities
- Conduct anti-bias testing and keep documentation as legal protection
- Enhance data security protocols for retained applicant information
- Reassess ROI of using AI tools, particularly for small businesses that may find the compliance burdens outweigh benefits
How This Impacts Employees
For job seekers and current employees, the updated regulation offers:
- Greater transparency and protection from discriminatory AI tools
- Improved fairness in resume screening, assessments, and interviews
- Legal recourse if AI-based systems are found to discriminate based on protected characteristics
Final Thoughts
California’s regulation is among the most comprehensive in the nation, moving beyond “fully automated” systems to include AI-assisted human decision-making. Employers must treat AI as a potential source of liability—and act accordingly to comply.
At DAVRON, we support ethical recruiting practices and technology that enhances, not replaces, human judgment. We’re here to help our clients stay ahead of compliance trends while ensuring top engineering, architecture, and construction talent is recruited fairly and transparently.