Many California employers using AI-powered hiring tools may not have reviewed the California Automated Decision Systems regulations. That is a problem, because the California Civil Rights Council rules have been in effect since October 1, 2025. They clarify that using an automated decision system can violate the Fair Employment and Housing Act (FEHA) when it discriminates against applicants or employees on a protected basis (California Civil Rights Council, June 30, 2025). Candidates who believe they were discriminated against can pursue complaints under that law.

This is not a future risk. It is a current one.

Who this applies to: Employers operating in California that use automated software to screen, score, rank, or otherwise evaluate job candidates may be covered. That can include AI resume screening tools, applicant tracking systems with automated scoring, and tools that produce a recommendation about a candidate. The rules also reach an employer's agents, meaning any person acting on behalf of an employer, which can include a screening vendor (Paul Hastings, August 12, 2025).

What the Regulation Says

The California Civil Rights Council regulations on automated-decision systems took effect October 1, 2025. They extend record retention to four years, apply to an employer's agents, and make evidence of anti-bias testing relevant to discrimination claims and defenses. Four practices follow from that, listed below. Only some of them are stated as mandates in the regulations, and each label says which.

Four Things to Have in Place

1
Records and Audit Trail (Required Retention) The regulations require employers to preserve personnel and other employment records, including automated-decision data, for four years (Paul Hastings, August 12, 2025). Keeping a log of every automated hiring decision, including who was evaluated, when, what the tool recommended, and what inputs it used, is a practical way to meet that duty and to respond to a regulator or an investigation.
2
Anti-Bias Testing (Not Mandated, but Relevant) The regulations do not mandate that employers test automated decision systems for bias, but "evidence, or lack of evidence, of anti-bias testing or similar proactive efforts" can bear on a discrimination claim or defense (Paul Hastings, August 12, 2025). The four-fifths rule is a common benchmark. If your tool consistently scores candidates from a protected class lower than a reference group, that can be evidence of disparate impact.
3
Candidate Transparency (Good Practice) We did not find a requirement in the California regulations that employers notify candidates when an automated tool is used. Telling candidates what role the tool plays in your process is nonetheless a sound transparency practice, and other laws may impose notice duties. Confirm the current requirements with employment counsel.
4
Documentation and Vendor Responsibility The regulations cover an employer's agents and contemplate liability for an AI vendor when an employer delegates an employment decision-making function to it (Paul Hastings, August 12, 2025). "We use a vendor" is not a complete answer. Expect to remain responsible for the tool's outputs regardless of who built it, and be ready to explain how the tool makes its recommendations.

A Common Mistake California Employers Make

The assumption that vendor liability covers employer liability. It may not.

If you subscribe to a resume screening platform and that platform produces biased outputs, a FEHA claim can be brought against your organization. The regulations also contemplate liability for a vendor that acts as your agent, so exposure may be shared, but shared exposure is not the same as shifted exposure. Any contractual indemnification is a separate matter between you and the vendor.

This means your compliance posture depends entirely on whether you can answer these questions:

If the answer to any of those is no, you may have a gap worth closing before a candidate or regulator asks.

Read the regulation directly: The California Civil Rights Council publishes the full ADS regulations at calcivilrights.ca.gov. If you are an HR professional or legal counsel, read the primary source rather than relying on summaries.

What the Four-Fifths Rule Means for Your Hiring Tool

The four-fifths rule appears in the federal Uniform Guidelines on Employee Selection Procedures (29 CFR 1607.4). In general, a selection rate for a group that is less than four-fifths (or eighty percent) of the rate for the group with the highest rate is regarded as evidence of adverse impact. It is a rule of thumb, not a safe harbor: the same guidelines note that smaller differences may still constitute adverse impact and that greater differences may not. California's regulations draw on the same federal guidelines, so the four-fifths test is a useful benchmark for California employers, but it is not the only test.

In a hypothetical example, if your AI tool recommends candidates from one demographic group for interviews at a rate of 60%, and recommends candidates from another group at a rate of 40%, that 67% ratio falls below the four-fifths threshold and can create legal exposure. Falling below four-fifths does not automatically prove discrimination, but the employer may then need to show that the tool is job-related and consistent with business necessity.

If you cannot produce documentation, that argument is much harder to make.

Why This Is Harder Than It Looks

Many AI hiring tools were not built with California ADS compliance in mind. Many generate a score and a recommendation without maintaining audit logs in a format that can be produced on regulatory request, running regular bias tests, or giving you an adverse impact report.

That means even if the underlying AI is not producing biased outputs, you have no way to prove it. And a tool that cannot show evidence of anti-bias testing is harder to defend than one that can.

The compliance infrastructure needs to be built into the tool from the start. Retroactive audits of tools without logging capabilities can be expensive and incomplete, and may be insufficient as evidence of ongoing compliance.

How TrueScan HR Addresses CA ADS Compliance

TrueScan HR was built with California ADS compliance as a design requirement. The compliance infrastructure runs at the platform level, in the background, without changing anything a hiring manager sees.

Audit logging captures every scan: timestamp, user ID, anonymized resume hash, job description hash, score, and recommendation. This log is retained and queryable on demand. If a candidate or regulator requests documentation of how their evaluation was handled, that record exists and can be produced.

Weekly automated bias testing runs 21 paired-resume comparisons every Monday across 9 protected dimensions: race and ethnicity, gender, age, pronouns, veteran status, disability, national origin, caregiving history, and intersectional combinations. Results are stored and retained for audit.

Proxy signal scrubbing strips signals that correlate with protected class membership but are not job-relevant, including graduation year as an age proxy and caregiving employment gaps as a gender proxy. This runs in the background without affecting the visible scan result.

Adverse impact reporting on Enterprise plans generates an EEOC 4/5ths analysis across your candidate pool on demand. A four-fifths analysis is a useful starting point for assessing whether a tool may be producing adverse impact.

If you use AI to screen candidates in California and your current tool cannot produce this documentation, see how TrueScan HR approaches compliance or read the California Civil Rights Council regulations directly.

California AI Hiring Guides

Practical guides to each piece of California's automated-decision rules for hiring:

Free download: the one-page California AI Hiring Compliance Checklist, with the regulation section behind every item. Get the checklist

Sources:

This post is general information, not legal advice. TrueScan HR is not a law firm. Confirm how these rules apply to your organization with employment counsel.