Can I use Ai to screen job applicants in California?

What does the rule say?

The California Civil Rights Council's automated decision system regulations sit under the Fair Employment and Housing Act, FEHA. They treat an automated system as part of the employment decision, not as a separate thing. If the system screens, scores, or ranks candidates, the rules apply to it.

The core idea is simple. Discrimination does not stop being discrimination because a machine did it. If your Ai screens out candidates in a way that harms a protected group, that can violate FEHA, even if nobody at your company intended it. Intent is not the test. Effect is.

That matters because most Ai hiring problems are not deliberate. A tool trained on past hiring data learns the patterns in that data, including the biased ones. The employer that bought the tool did not intend to discriminate. The law still holds the employer responsible.

What counts as an automated decision system?

An automated decision system is any system that uses Ai to screen, score, rank, or select candidates, or to make or help make an employment decision. That covers resume screeners, video interview scorers, assessment tools that rank applicants, and chatbots that filter candidates before a human sees them.

It does not cover a spreadsheet that sorts by years of experience. The rule reaches systems that make judgments, not tools that organize information. If the tool decides who advances, it counts.

Does the vendor take the liability?

This is the part most employers miss. Buying the tool does not buy your way out of the rule. The employer stays liable even if a vendor built the tool, trained the model, or promised it was fair. A vendor contract that says the tool is compliant is not a defense under FEHA.

That does not mean you cannot use a vendor. It means you have to do your own diligence. Ask the vendor for bias testing. Ask what data the tool uses. Ask how it was validated. Get the answers in writing, and keep them with your records.

How long do I keep records?

Keep related records for four years. That covers the tool's outputs, the decisions made, the vendor's testing documentation, and the notices given to candidates. Four years is longer than most federal retention rules, so California sets the timeline for any employer that hires in the state. Use the longest timeline that applies.

What do I do first?

Four steps, in order.

1. List every tool that screens, scores, or ranks candidates. Include the ones your recruiters use without telling you. If you do not know the full list, you cannot comply with the rule.
2. Ask each vendor for bias testing and what data the tool uses. Get it in writing. If a vendor cannot or will not answer, that is information about the vendor.
3. Tell candidates when Ai is involved. Notice is the piece most employers skip, and it is the easiest to fix. A sentence in the posting and a sentence in the application flow is a good start. Confirm the exact wording with counsel.
4. Keep the records. Four years, as above, and keep them organized enough to find.

How does California compare?

California is the broadest of the state Ai hiring laws. It covers automated decisions across employment, not only hiring, and it holds the employer liable for disparate impact. Illinois reaches similar ground through its Human Rights Act. Texas reaches only intentional discrimination. New York City requires a bias audit. California does not require a standalone bias audit. Colorado's replacement law is aimed at January 2027.

If you hire in more than one state, build to the strictest standard in the set. For most employers, that is California.

HR guidance, not legal advice.

What to do next

Read why Ai needs a strategy for the framework behind the rules, then check the FAQ below for the short answers. If you want someone to map your tools against the California rules, that is what an Ai risk and compliance audit is for.