WThe Warden WatchTracking AI Compliance in HR

The Warden Watch

Tracking AI Compliance in HR

A public monitor for regulation, litigation and incidents affecting AI in hiring, recruitment and employment decision-making.

Regulation intelligence node

Employment Regulations Regarding Automated-Decision Systems

California FEHA Regulations Regarding Automated-Decision Systems

PassedBoth

Jurisdiction

California, United States

Status

Passed

Live from

1 October 2025

Compliance deadline

1 October 2025

What this regulation is

These regulations clarify how California's workplace protections under the Fair Employment and Housing Act (FEHA) apply to automated-decision systems (ADS), machine learning, and artificial intelligence[cite: 1, 2]. Deploying automated software to screen resumes, target job advertisements, rank candidates, or grade online interviews is unlawful if it generates an unjustified adverse impact against protected groups or relies on discriminatory proxy variables[cite: 1]. Furthermore, any third-party software provider or recruitment vendor acting as an 'agent' of the employer to execute traditional HR screening functions is legally classified as an employer and faces direct statutory liability[cite: 1].

Regulator or body

California Civil Rights Council

Audience impact

What different teams need to know

HR and recruitment teams

HR departments deploying automated vendor systems, parsing software, or online psychometric tools must monitor their selection criteria to prevent systemic adverse impact[cite: 1]. Employers must provide immediate alternative testing pathways and reasonable accommodations if an automated challenge model (e.g., dexterity or reaction time games) disrupts accessibility for neurodivergent individuals, disabled applicants, or specific religious observances[cite: 1]. Additionally, internal personnel record-keeping policies must be updated to extend data-retention schedules for all automated evaluation metrics, candidate inputs, and scoring outputs from two to four years[cite: 1].

HR technology vendors

Software developers, recruitment platform providers, and automated assessment intermediaries are legally classified as 'agents' of an employer under FEHA if they execute screening, matching, or recommendation tasks[cite: 1]. This classifies vendors as employers under California law, exposing them to direct liability[cite: 1]. Vendors must re-architect data infrastructure to support a four-year data-preservation window and maintain clear testing logs, as a client's compliance defence will depend heavily on the presence and quality of proactive anti-bias validation records[cite: 1].

Status intelligence

Current status and key dates

Regulatory status

Approved by the Office of Administrative Law on 27 June 2025; coming into force on 1 October 2025[cite: 2].

Date timeline

Passed
27 June 2025
Live from
1 October 2025
Enforcement / compliance
1 October 2025

Mandatory audit scope

Bias auditing and protected characteristics

Bias audit summary

The source mentions anti-bias testing as evidence that may be considered, but does not appear to create a freestanding external, independent, periodic or pre-deployment bias audit requirement[cite: 1]. Proactive anti-bias validation records are established as an evidentiary factor to evaluate the quality, scope, and response of an employer or vendor when defending or establishing a claim of discrimination[cite: 1].

Protected characteristics summary

Protections apply recursively across all categories safeguarded by FEHA, explicitly referencing race, national origin (including linguistic accents, English proficiency, and immigration status), sex, gender, pregnancy or perceived pregnancy, marital status, religious creed, age (40 or older), physical or mental disabilities (including uncorrected vision or hearing, and past drug addiction), and pre-offer criminal history filters[cite: 1].

Duties and legal nuance

Obligations and requirement levels

Binding duties are shown as compact cards. Evidence and notes are collapsed by default.

Recordkeeping

BindingApplies to: BothExpress
Trigger and timing

Trigger: Creation or receipt of employment records, applications, or automated-decision system data dealing with employment practices[cite: 1].

Timing: Preserved for a period of four years from the date of creation of the record or the personnel action involved, whichever is later[cite: 1].

Source: § 11013(c)

Evidence
Any personnel or other employment records made or kept created or received by any employer or other covered entity dealing with any employment practice and affecting any employment benefit of any applicant or employee (including all applications, personnel, membership or employment referral records or files) shall be preserved by the employer or other covered entity for a period of twefour years from the date of the making of the record or the date of the personnel action involved, whichever occurs later. This includes all applications, personnel records, membership records, employment referral records, selection criteria, automated-decision system data, and other records created or received by the employer or other covered entity dealing with any employment practice and affecting any employment benefit of any applicant or employee[cite: 1].
Notes
This extends the previous standard California personnel retention mandate from two years to four years, explicitly incorporating automated-decision system data, dataset inputs, and system outputs[cite: 1].

Bias Audit

MentionedApplies to: BothExpress
Trigger and timing

Trigger: When evaluating, reviewing, or defending an employment selection claim or affirmative defence under the Act[cite: 1].

Timing: Proactive context[cite: 1].

Source: § 11009(f)

Evidence
Relevant to any such claim or available defense is evidence, or the lack of evidence, of anti-bias testing or similar proactive efforts to avoid unlawful discrimination, including the quality, efficacy, recency, and scope of such effort, the results of such testing or other effort, and the response to the results[cite: 1].
Notes
The source mentions anti-bias testing as evidence that may be considered, but does not appear to create a freestanding external, independent, periodic or pre-deployment bias audit requirement[cite: 1]. This evidentiary tracking expectation is mirrored in sections § 11020(b)(i) and § 11028(m)[cite: 1].

Accessibility Accommodation

BindingApplies to: DeployerExpress
Trigger and timing

Trigger: Use of an automated-decision system that screens, ranks, or measures an applicant's skills, dexterity, reaction time, speech patterns, facial expressions, or physical traits[cite: 1].

Timing: Prior to or during test administration and selection stages[cite: 1].

Source: § 11016(c)(5), § 11016(d)(1)

Evidence
The use of an automated-decision system that, for example, measures an applicant's skill, dexterity, reaction time, and/or other abilities or characteristics may discriminate against individuals with certain disabilities or other characteristics protected under the Act. To avoid unlawful discrimination, an employer or other covered entity may need to provide reasonable accommodation to an applicant as required by Article 8 (religious creed) or Article 9 (disability) of these regulations[cite: 1].
Notes
Confirms that automated evaluation systems or algorithmic puzzles cannot run without accommodating candidates[cite: 1]. Section 11072(b)(5)(F) specifies that the use of an automated-decision system, in the absence of additional process or actions, does not constitute an individualized assessment[cite: 1].

Actor split

Who the record says is responsible

Primary liability role: Both

Employer / employment agency duties

Recordkeeping · BindingBias Audit · MentionedAccessibility Accommodation · Binding

Vendor support context

Software developers, recruitment platform providers, and automated assessment intermediaries are legally classified as 'agents' of an employer under FEHA if they execute screening, matching, or recommendation tasks[cite: 1]. This classifies vendors as employers under California law, exposing them to direct liability[cite: 1]. Vendors must re-architect data infrastructure to support a four-year data-preservation window and maintain clear testing logs, as a client's compliance defence will depend heavily on the presence and quality of proactive anti-bias validation records[cite: 1].

Hiring and employment use cases

Affected use cases

SourcingExpressly coveredExpress

Directing job advertisements or other recruiting materials to targeted groups[cite: 1].

Source: § 11008.1(a)(1)(B)

Targets algorithmic demographic ad-delivery networks and programmatic social sourcing configurations[cite: 1].
ScreeningExpressly coveredExpress

Screening resumes for particular terms or patterns[cite: 1].

Source: § 11008.1(a)(1)(C)

Encompasses automated cv filtering platforms, semantic matches, and keyword parsers[cite: 1].
InterviewingExpressly coveredExpress

Analyzing facial expression, word choice, and/or voice in online interviews[cite: 1].

Source: § 11008.1(a)(1)(D)

Covers asynchronous AI video interview tracking tools, NLP speech analytics, and computer-vision behavioral metrics[cite: 1].
AssessmentExpressly coveredExpress

Using computer-based assessments or tests, such as questions, puzzles, games, or other challenges[cite: 1].

Source: § 11008.1(a)(1)(A)

Includes gamified behavioral screening configurations and psychometric challenges[cite: 1].
Background ChecksExpressly coveredExpress

inquiring about criminal history through an employment application, background check, er internet searches, or the use of an automated-decision system[cite: 1].

Source: § 11017.1(a)(1)

Bans using an ADS or background screening bot to evaluate or weed out criminal data before a conditional offer is extended[cite: 1].
Workforce AnalyticsExpressly coveredExpress

Screen, evaluate, categorize, and/or recommend applicants or employees[cite: 1].

Source: § 11008.1(a)(1)(A)(iv)

Includes predictive internal mobility models, scoring tools for workforce distributions, and algorithmic personnel trackers[cite: 1].
PromotionExpressly coveredExpress

which may include applicant recruitment, applicant screening, hiring, promotion, or decisions regarding pay, benefits, or leave[cite: 1].

Source: § 11008(a)

Primary source

Source

Regulator or body

California Civil Rights Council

Warden Watch alerts

Track AI hiring compliance changes

Get weekly alerts on AI hiring regulation, litigation and incidents.

Employment Regulations Regarding Automated-Decision Systems — The Warden Watch