California AI workplace laws: AB 1883, SB 951, SB 947 explained
A client with offices in Los Angeles called us two weeks ago. Her head of HR wanted to replace the annual review process with an “AI performance coach” that would score employee emails, camera-on meetings, and Slack sentiment. The vendor pitch arrived the same week Governor Gavin Newsom signed three new laws on September 30, 2026. The pitch is now unusable in California without serious redesign. The California AI workplace laws — AB 1883, SB 951, and SB 947 — reach every private employer in the state and take effect on January 1, 2027, with civil penalties on the AI surveillance side and new disclosure obligations on the layoff side. This post walks through what each bill actually says, how it interacts with the existing WARN Act in California Labor Code §§1400–1408, and the compliance checklist we are now giving clients before the new year.
What does AB 1883 ban about AI workplace surveillance? 🧠
AB 1883 prohibits a California employer from using an AI-powered “workplace surveillance tool” to recognize, infer, or predict an employee’s emotional state, or to collect that employee’s neural data, even if the information is never used to make an employment decision.
The bill’s definitions do most of the work. A “workplace surveillance tool” is any system that collects employee data, communications, biometrics, or behavioral information other than through direct human observation. That reaches further than most employers assume: video and audio monitoring, time-and-attendance platforms, geolocation on company devices, software that scores email or chat tone, and even call-center quality-assurance tools that flag “frustration.” If any of these tries to recognize, infer, or predict how an employee is feeling, AB 1883 shuts it off at the California payroll population.
“Neural data” is defined by reference to activity of an individual’s central or peripheral nervous system — the kind of information generated by emerging consumer neurotechnology devices and headsets. Collecting that information from employees is prohibited outright. The prohibition applies even where the data never leaves the vendor’s servers and even where no manager ever sees it. Civil penalties run up to $500 per violation, and the law covers employees but not applicants.
There is a narrow safety exception for tools used to ensure workplace safety. The scope of that exception is not spelled out in the bill, and the practical read for now is: a wearable that detects fatigue in a heavy-equipment operator may fit; a call-center emotion scorer dressed up as a “wellness” tool almost certainly does not.
What are SB 951’s new AI-related WARN Act disclosures? ⚠️
SB 951 amends the existing California WARN Act in Labor Code §§1400–1408 to require extra disclosures when a covered mass layoff, relocation, or termination is caused “in whole or in substantial part” by AI or other automated technology.
California’s WARN Act already requires 60 days’ written notice to affected employees, the Employment Development Department, and local workforce agencies before a covered action. Covered actions include mass layoffs of 50 or more employees in a 30-day period, relocations of 100 or more miles, and plant terminations. The California EDD WARN Act portal houses the current notice forms and tracks filings publicly. SB 951 adds a technology-displacement overlay on top of that machinery.
When the triggering action is caused by automation in whole or in substantial part, the notice now must include:
- A statement at the top of the notice: “This notice is for a technology displacement.”
- The number, classification or occupation, and work location of the affected positions.
- The specific job functions being automated.
- The category or type of technology responsible for the displacement.
The phrase “in substantial part” is where the practical work lives. Reductions in force rarely have a single cause. A warehouse that automates a packing line and also cuts a shift because of a demand drop may need the new disclosure for part of the layoff but not the whole of it. The bill does not supply a bright-line test, so expect enforcement to be fact-driven and expect labor-side counsel to argue for a broad reading.
WARN Act monitoring is now an AI question too
Employers that have treated WARN monitoring as a headcount spreadsheet should widen it. The new technology-displacement disclosure requires employers to know, before the notice is drafted, which positions were affected by automation and what the underlying technology does. In our practice we are rebuilding WARN tracking so that it captures the business reason for every eliminated role — the forecast failure, the process-automation project, the model migration — alongside the headcount math.
When does SB 947 require human review of AI employment decisions? 👥
SB 947 bars California employers from relying primarily on an AI or algorithmic decision system to discipline or terminate an employee. A human reviewer must independently corroborate the system’s output using relevant evidence before the employment decision is made, and the employee must receive a standalone written notice.
The bill imposes three operational obligations:
- Human corroboration. The reviewer must independently assess the AI output against supervisory evaluations, personnel files, work product, or other relevant evidence. A manager who rubber-stamps the algorithm is not corroboration — the record has to show that the human engaged with the underlying facts.
- Standalone written notice. When the discipline or termination is communicated to the employee, the employer must deliver a separate written notice that identifies the automated decision system in use, explains the human-review step, and describes the employee’s rights under the law.
- Data access right. Employees may request a description of the data about themselves that the system used during the preceding 12 months. Employers need a process to produce that description on request without triggering a separate privacy or discovery problem.
| Bill | What it does | Covered employers | Effective date |
|---|---|---|---|
| AB 1883 | Bans AI-powered workplace surveillance that infers emotional state or collects neural data; up to $500 per violation. | All private employers in California (employees, not applicants). | January 1, 2027 |
| SB 951 | Adds technology-displacement disclosures to California WARN Act notices when AI or automation causes the layoff in whole or substantial part. | Employers covered by the existing California WARN Act (generally 75+ employees). | January 1, 2027 |
| SB 947 | Requires human review and a standalone written notice before an AI or algorithmic decision system is used to discipline or terminate an employee. | California employers using automated decision systems in employment actions. | January 1, 2027 |
From our practice: the hidden cost is documentation, not technology
In our practice, the compliance cost of these laws is not buying different software — it is proving that a human was actually in the loop. The clients that will struggle in a wage-and-hour or wrongful-termination case are the ones whose HRIS shows a manager clicking “approve” four seconds after the algorithm’s flag arrived. We are advising clients to design the human-review step so that the reviewer has to attach a short written rationale — a sentence tying the AI output to a specific piece of corroborating evidence — before the system will record the decision. That artifact is what defends SB 947 compliance later.
What should California employers do before January 1, 2027? 📅
Inventory every AI, algorithmic, and surveillance tool touching California employees; rewrite WARN Act monitoring to flag automation as a cause; redesign disciplinary and termination workflows around documented human review; and train managers before the first 2027 action is initiated.
- AI and surveillance inventory. List every tool that collects employee data, scores communications, measures productivity, monitors cameras or audio, tracks location, or makes or informs employment decisions. For each, record what the vendor says it does, what it actually does, where it runs, and who has access to its outputs.
- Emotion and neural-data audit. Flag any tool that claims to measure mood, stress, engagement, tone, sentiment, or focus from employee signals. Decide whether to turn it off for California employees, limit the feature, or seek narrow-safety-exception legal advice before January 1.
- WARN Act plumbing. Update the trigger template to capture whether automation contributed to each eliminated role. Pre-draft the technology-displacement notice language so legal review is faster when it is needed. Confirm the EDD WARN filing workflow still works.
- Human-review workflow. Build the standalone written notice. Build the process by which a reviewer corroborates AI output with independent evidence and leaves a written trail. Build the response process for the 12-month data-description request.
- Manager training. Train before the first 2027 discipline decision, not after. The civil-penalty exposure is on the employer, but the daily operational risk sits with the manager who approves the algorithm’s recommendation without reading it.
- Vendor contract review. Push vendors for written confirmation that their tool does not infer emotional state or collect neural data for California users, and for an export of the 12-month data description employees can now request under SB 947.
For employers below the WARN Act headcount threshold, SB 951 does not apply, but AB 1883 and SB 947 do. A ten-person startup using an off-the-shelf AI performance tool has the same exposure under AB 1883 and SB 947 as a 10,000-person employer. The civil penalty structure makes small employers particularly sensitive to repeated violations across a California team.
Summary: California AI workplace laws
- Three laws signed September 30, 2026, effective January 1, 2027 — AB 1883, SB 951, and SB 947.
- AB 1883 bans AI workplace surveillance that infers emotion or collects neural data; up to $500 per violation; narrow safety exception; all private employers.
- SB 951 adds a technology-displacement disclosure to California WARN Act notices when AI or automation causes a layoff in whole or in substantial part.
- SB 947 requires documented human review and a standalone written employee notice before an AI or algorithmic system is used to discipline or terminate a worker.
- Compliance is mostly documentation: an AI inventory, revised WARN tracking, a human-review workflow with written rationale, and manager training before the first 2027 action.
Frequently asked questions about California AI workplace laws ❓
Q. When do the new California AI workplace laws take effect?
Most provisions of the three new California AI workplace laws take effect on January 1, 2027. AB 1883, SB 951, and SB 947 were signed by Governor Gavin Newsom on September 30, 2026, giving employers roughly three months to review their AI tools, surveillance practices, WARN Act procedures, and disciplinary policies before enforcement begins.
Q. Does AB 1883 apply to small employers in California?
Yes. AB 1883’s prohibition on AI-powered workplace surveillance that infers employee emotional state or collects neural data applies to private employers of all sizes. The law covers employees but not job applicants, and civil penalties can reach $500 per violation. There is a narrow exception for tools used to ensure safety, but the scope of that exception is not spelled out in the bill.
Q. What triggers an SB 951 technology-displacement WARN notice?
SB 951 attaches to the existing California WARN Act in Labor Code §§1400–1408 and requires an added technology-displacement disclosure when a covered mass layoff, relocation, or termination is caused “in whole or in substantial part” by AI or other automated technology. The notice must state at the top that it is for a technology displacement and must list the affected positions, the job functions being automated, and the category of technology responsible.
Q. Can California employers still use AI to screen or discipline employees?
Yes, but under SB 947 an employer cannot rely solely on an AI or algorithmic decision system when disciplining or terminating a worker. A human reviewer must independently corroborate the system’s output using relevant evidence such as supervisor evaluations, personnel files, or work product, and the employee must receive a standalone written notice explaining the use of the system and their rights under the law.
Q. How should California employers prepare for January 1, 2027?
Start with an inventory of every AI, algorithmic, and surveillance tool in use, mapped to the people and decisions it touches. Update WARN Act monitoring to flag automation as a potential cause of workforce reductions. Rewrite disciplinary and termination procedures to include a documented human-review step and a written employee notice. Train managers on the new rules and brief payroll, HR, and finance teams on the civil-penalty exposure before any 2027 layoff round.
Q. Do the California AI workplace laws apply to remote employees working outside California?
The laws protect workers whose employment relationship is governed by California law, which generally includes California-based employees working remotely for a California employer. Employers with multistate workforces should expect their California payroll population to be covered and should assume state-by-state variation; other states are watching the California model and may follow with their own rules.
This article is general information, not legal advice for a specific situation. California AI employment compliance interacts with federal labor law, existing WARN Act procedures, and the California Consumer Privacy Act. If you operate in California and are unsure how AB 1883, SB 951, or SB 947 reaches your workforce, contact SW Accounting & Consulting Corp for a confidential review before January 1, 2027. See California Legislative Information for the full text of each bill.
