Go to integrated search

California AI Employment Law: Termination Rules and Effective Dates

Jurisdiction:California

California AI employment law sets different rules for automated termination decisions and job cuts caused by AI replacing workers.

Some protections already apply; SB 947’s human review and notice requirements begin July 1, 2027. Employees and employers need to distinguish AI-assisted discipline from technology-driven layoffs. This guide explains which rules apply, when they take effect, and what records help assess a workplace decision.



1. Which Rules Apply before and after 2027?


Start with the date and type of employment decision. Existing discrimination protections already cover automated tools, while the new laws add requirements on separate schedules. Signing a law does not necessarily make its workplace obligations immediately enforceable.


Three Frameworks with Different Starting Dates

Legal FrameworkEffective or Operative DatePrimary Focus
FEHA automated-decision regulationsOctober 1, 2025Discrimination involving automated employment decisions
SB 951January 1, 2027Technology-displacement disclosures in qualifying Cal-WARN notices
SB 947July 1, 2027Automated discipline and termination, human review, and employee notice

FEHA automated-decision regulations

  • Effective or Operative DateOctober 1, 2025
  • Primary FocusDiscrimination involving automated employment decisions

SB 951

  • Effective or Operative DateJanuary 1, 2027
  • Primary FocusTechnology-displacement disclosures in qualifying Cal-WARN notices

SB 947

  • Effective or Operative DateJuly 1, 2027
  • Primary FocusAutomated discipline and termination, human review, and employee notice

These are California state requirements. Federal employment discrimination laws and the federal WARN Act may also apply, but each has separate coverage rules.

Existing Rights Remain Relevant in 2026

The Fair Employment and Housing Act, or FEHA, already prohibits covered employment discrimination. Its automated-decision regulations explain how those protections apply when software makes or assists decisions affecting applicants and employees.

A worker who receives an allegedly biased evaluation in 2026 should assess existing employment discrimination protections rather than wait for SB 947. The employer’s use of software does not eliminate its responsibility, and a manager’s approval does not automatically resolve a discriminatory outcome.


2. When Does Workplace Software Qualify As an ADS?


The relevant question is what the system does, not whether its vendor calls it AI. A tool that recommends an outcome can affect an employment decision even when a person formally makes the final choice.


Scores and Recommendations Can Fall within the Definition

SB 947 defines an automated decision system, or ADS, to include specified computational processes that produce simplified outputs, such as scores, classifications, or recommendations. Those outputs must assist or replace human discretionary decision-making and materially affect people.

For example, a performance score that substantially drives a disciplinary recommendation deserves closer examination. Ordinary databases, calculators, spam filters, and specified security tools fall outside SB 947’s ADS definition.

Identify the Decision before Selecting the Legal Framework

The FEHA regulations address automated systems across employment decisions, including applicant screening. SB 947’s specific human-review and postuse-notice requirements focus on discipline and termination.

The frameworks use different definitions and serve different purposes. An automated hiring rejection, an employee warning, and a position eliminated through automation should not receive the same legal analysis simply because each involves technology.


3. Human Review and Employee Notice under SB 947


Diagram: For discipline or termination primarily based on ADS output, a human checks supporting facts; unusable output is excluded, and the employee receives notice of the decision.
Diagram: For discipline or termination primarily based on ADS output, a human checks supporting facts; unusable output is excluded, and the employee receives notice of the decision.

Beginning July 1, 2027, SB 947 prohibits employers from relying solely on an ADS when making disciplinary or termination decisions. When an employer primarily relies on automated output, the law also requires corroboration and notice.


A Human Must Check the Supporting Facts

The reviewer must corroborate the decision using data collected or used to produce the output, or other relevant supporting information. That material may include managerial evaluations, personnel files, work product, peer reviews, or witness interviews.

If the employer cannot corroborate the output, it cannot use that output for the decision. The same restriction applies when the reviewer concludes that the output is inaccurate, incomplete, or misleading.

Employers preparing for the operative date should build a process that records what the reviewer checked. A signature alone does not explain whether anyone verified the facts behind a score.

Notice Explains the Automated Reliance

When an employer primarily relies on an ADS, it must give the affected employee a separate, plain-language written notice when communicating the decision. The notice must address:

  • The employer’s primary reliance on an ADS.
  • Human review and corroboration of the output.
  • A human contact and the employee’s data-description rights.
  • Protection against retaliation for exercising statutory rights.

The employee may request a meaningful, objective description of their own data used by the system. That right does not necessarily provide source code or other employees’ personal information.

These requirements explain how a decision occurred. Whether it also supports a wrongful termination claim depends on the applicable legal grounds and surrounding facts.


4. AI Job Replacement and Cal-WARN Notice Requirements


AI can help select a worker for dismissal, or it can replace the work itself. SB 951 addresses technology-driven displacement by adding information to notices required under Cal-WARN. It does not prohibit employers from automating positions.


Determine Whether Cal-WARN Covers the Event

A covered establishment generally includes a qualifying industrial or commercial facility that employs, or employed within the preceding 12 months, at least 75 people. A mass layoff generally involves at least 50 employees at a covered establishment within a 30-day period.

Cal-WARN also covers qualifying relocations and operational terminations. In this statute, “termination” means cessation or substantial cessation of operations, rather than every individual firing. Employee eligibility and statutory exceptions also matter.

Federal WARN requires a separate analysis. Meeting or missing one law’s threshold does not settle coverage under the other.

Disclose the Functions That Technology Will Replace

Starting January 1, 2027, SB 951 adds disclosures for covered events caused wholly or substantially by AI or other automated technology replacing or automating positions.

The notice must identify affected layoffs by number, occupation or classification, and location. It must also describe the functions that technology will automate and the category or type of technology causing the displacement. The prescribed technology-displacement statement belongs at the top of the notice.

The existing 60-day advance-notice framework remains central, subject to applicable exceptions. SB 951 does not require 60 days’ notice for every dismissal involving AI.


5. Frequently Asked Questions


These questions address exceptions and overlapping obligations that can change the analysis beyond the main review and notice requirements.


No. The exception requires a qualifying collective bargaining agreement with a clear, explicit waiver and specified provisions, including protection from algorithmic management. Union representation alone does not establish the exception.

SB 951 directs the Employment Development Department to study technology-related hiring changes and report to the Legislature. That provision does not establish a blanket employer notice requirement whenever a business stops hiring for a role.

No. SB 947 preserves applicable privacy-related automated-decision regulations for businesses subject to the California Consumer Privacy Act. Completing an employment notice does not, by itself, satisfy separate privacy obligations.


6. Assessing a Decision or Preparing a Review Process


The most useful records connect the system’s output to the actual decision. Employees need to understand what drove the outcome. Employers need to explain what a reviewer checked and why the supporting facts justified the action.


Preserve the Decision Trail

Relevant records may include automated scores, performance reports, correction requests, human evaluations, and written notices. Employees should preserve documents they lawfully possess. Employers should identify the tool’s settings, data sources, and review history before records change or disappear.

A discrepancy deserves attention. A low score based on missing information, for instance, may require a different assessment from a score supported by documented performance problems.

Match the Concern to the Available Legal Route

SB 947 provides enforcement through the Labor Commissioner and public prosecutors. Employees should not assume that it creates a standalone private lawsuit for every violation. Existing discrimination, retaliation, or other claims require their own analysis and deadlines.

For planned deployment or workforce changes, employment counseling can help align review procedures and notices with the applicable dates. For a disputed decision, bring the decision notice, available scoring records, and a timeline so an attorney can identify the relevant protections and enforcement options.

05 Oct, 2026


The information provided in this article is for general informational purposes only and does not constitute legal advice. Prior results do not guarantee a similar outcome. Reading or relying on the contents of this article does not create an attorney-client relationship with our firm. For advice regarding your specific situation, please consult a qualified attorney licensed in your jurisdiction.
Certain informational content on this website may utilize technology-assisted drafting tools and is subject to attorney review.

Online Consultation
Phone Consultation