California Pumps the Brakes on Artificial Intelligence

updated on April 24, 2025

Remember “Rosie the Robot” from The Jetsons? While the popular show aired in the 1960s, it was set in 2062, with a plethora of futuristic advancements from flying cars and holograms, to helpful robots, like Rosie. Rosie took care of household chores such as cooking, cleaning, laundry, and childcare. With advancements in Artificial Intelligence (AI), it’s not difficult to imagine a world like this in another 40 years!

However, there is a downside to technological advancement, particularly when it is misused, even inadvertently. California’s Civil Rights Department (CRD) has recognized these risks and at its March 21, 2025 meeting, approved final regulations aimed at addressing potential discrimination when AI tools are used to make personnel decisions, such as screening applicants. This includes any “computational process that makes a decision or facilitates human decision making . . . [through] artificial intelligence, machine learning, statistics, and/or other data processing techniques.” The regulations must be approved by the Office of Administrative Law and are expected to go into effect later this year.

While AI tools have gained massive popularity for employment, California employers should proceed cautiously based on the CRD developments.

CRD’s Prohibitions on Automated-Decision/AI Tools

Key highlights from the CRD’s regulations include:

• FEHA applies to AI: The regulations clarify that California’s antidiscrimination laws (e.g., FEHA) and regulations apply to discrimination caused by the use of artificial intelligence/automated-decision systems. Both employers as well as third-parties that sell automated decision systems could be held liable;
• AI tools may result in “adverse-impact” discrimination: Employers may be liable even if they did not intentionally use AI to discriminate. For example, an employer and its agents would be liable if the AI tool resulted in an “adverse impact” on a protected group (e.g., disproportionately screened out a certain race, gender, etc.), when it is not job related and consistent with business necessity;
• AI alone is insufficient for individualized assessments: California employers with 5 or more employees are required to conduct an “individualized assessment,” assessing certain factors, before denying an applicant a position based on their conviction history. The regulations clarify that using an automated decision system alone is insufficient to satisfy the individualized assessment process.
• AI cannot be used for medical or psychological inquiries: The regulations clarify that most medical or psychological exams or inquiries are still prohibited when using an automated decision system.
• Record retention is required: The regulations require employers to hold on to AI-related records for at least four years. This includes application, selection criteria, automated decision system data, and other records created or received by the employer that affected any employment benefit, applicant, or employee.

What Else Should Employers Consider?

Aside from proceeding with caution in using AI tools for hiring and promotion decisions, employers should also consider whether—and to what extent—they want employees using AI to carry out their job duties. This can raise questions regarding confidentiality issues and protecting proprietary information, the quality and accuracy of employee work products, and more. CEA members can get started with our downloadable Sample AI Policy on our HR Forms page. Give us a call with questions at 800-399-5331.