Skip to content
HRaizon Subscribe

AB 1883 Would Ban Two Uses of Workplace AI—If It Becomes Law

California AB 1883 would restrict workplace AI emotion recognition and neural-data collection. Status and practical scope as of September 29, 2026.

Share X in f
Priya Ellison

Status checked September 29, 2026: California AB 1883 has passed the Legislature, but it is not yet law. The official bill history says it was presented to Governor Gavin Newsom on September 10 and records no later signature or veto. September 30 is the deadline for the governor to sign or veto bills in this category, according to the California Senate’s 2026 legislative calendar. If the governor does not return the bill by that date, it becomes a statute without his signature under Article IV, Section 10 of the California Constitution.

If enacted, AB 1883 would prohibit two specific uses of AI-based workplace surveillance on employees: emotion recognition and neural-data collection. It would not ban workplace surveillance generally.

What AB 1883 would prohibit

The final enrolled bill text says an employer could not use a workplace surveillance tool that uses artificial intelligence on employees to:

  1. Recognize an individual’s emotional state, or make inferences or predictions about it.
  2. Collect neural data.

“Neural data” means information generated by measuring activity in an employee’s central or peripheral nervous system, excluding information inferred from nonneural sources.

The emotion restriction is not limited to brain or nerve measurements. A tool could potentially fall within it if, for example, it processes an employee’s face, voice or behavior and produces an employee-level emotion inference or prediction. Relevant questions include what the tool collects, what its AI produces, whether it is used on employees and whether an exception applies—not whether a vendor calls the product sentiment analysis, engagement monitoring, coaching or wellness software.

The definition of “workplace surveillance tool” is broad. It covers systems that collect or facilitate the collection of employee data, activities, communications, actions, biometrics or behaviors by means other than direct human observation. The bill lists video and audio surveillance, continuous incremental time tracking, geolocation and photo-optical systems as examples.

What the bill would not ban

AB 1883 is narrower than some descriptions of earlier versions suggest.

Question Final enrolled text
Does it ban all employee monitoring? No. Monitoring that does not perform either prohibited function is not banned by AB 1883.
Does it generally ban facial recognition? No. The final text does not contain the earlier version’s facial-recognition restriction.
Does it ban gait analysis or inferences about protected activity? Not expressly. Those provisions appeared in an April committee analysis but are absent from the enrolled bill.
Does it cover ordinary human observation? No. The surveillance-tool definition concerns collection by means other than direct observation by a person.
Does it expressly cover job applicants? No. The operative prohibition applies to “employees,” defined as people employed by an employer.

That applicant distinction matters for recruiting teams. Based on the enrolled language, an emotion-scoring interview product used only on external candidates does not appear to be within AB 1883’s employee-focused prohibition. A current employee applying internally would still be an employee, however. And being outside AB 1883 would not establish that a product is lawful: discrimination, disability, privacy, biometric and local automated-hiring rules may still apply. The same product could also enter AB 1883’s scope if used in onboarding, training, performance management or discipline.

AB 1883 says it does not prohibit workplace surveillance used to ensure safety. The enrolled text does not define the limits of that clause, so an employer should not assume that a general vendor claim that a tool “improves safety” settles the issue. Separate, narrow exemptions cover surveillance related to aircraft development and the development of products or services for national-security, military, space or defense purposes. Those exemptions apply only where the surveillance is reasonably necessary to comply with specified federal authority or a binding federal contract.

Which employers would be covered?

The bill contains no employee-count threshold. Its definition reaches private employers and governmental entities, including all branches of state government, the University of California, California State University, cities, counties, school districts and other state or local entities.

It also expressly includes labor contractors, farm and foreign labor contractors, client employers and certain entities or individuals contracting for labor or property services. Organizations therefore need to examine tools deployed through staffing and labor-supply arrangements, not just software purchased directly by HR.

Enforcement and possible effective date

The bill would authorize enforcement by the Labor Commissioner and public prosecutors. Available measures include investigations, citations, civil actions and temporary relief. A violation could carry a penalty of up to $500 per violation; the text prevents recovery of both a statutory penalty paid to an employee and a civil penalty for the same violation. Local ordinances offering equal or greater employee protection would not be preempted.

AB 1883 contains no special effective date or urgency clause. If it is signed or allowed to become law by the September 30 deadline, it would ordinarily take effect January 1, 2027 under Article IV, Section 8 of the California Constitution. Until then, HR teams should describe it as an enrolled bill—not a current California ban.

What HR teams can check now

  1. Inventory functions, not product names. Search documentation and settings for emotion, affect, sentiment, mood, stress, attention, empathy, engagement, fatigue and neural or nervous-system data.
  2. Record the actual workflow. Identify the inputs, outputs, affected people, business purpose, recipients and whether a score influences coaching, scheduling, evaluation, discipline or termination.
  3. Separate applicants from employees. Record whether a person is an external applicant, a current employee applying internally or an employee subject to post-hire assessment.
  4. Ask vendors for written answers. Confirm whether the system recognizes, infers or predicts emotion; whether it measures central or peripheral nervous-system activity; and whether those functions can be disabled and verified as disabled.
  5. Test claimed exceptions cautiously. Document the specific safety purpose and how the tool is configured rather than relying on broad marketing language.
  6. Review connected decisions. Turning off an emotion label does not resolve separate concerns when another AI score drives discipline or dismissal. California’s enrolled No Robo Bosses Act addresses a different mechanism: automated employee discipline and termination.
  7. Watch the official bill history. A signature, veto or expiration of the veto deadline will determine AB 1883’s status.

For employees, a useful first step is to ask what data a monitoring tool collects, what it infers, who receives the output and whether that output affects employment decisions. Preserve the employer’s notice, screenshots and product name. Whether a particular deployment violates current law—or would violate AB 1883—depends on how it operates and whom it is used on.

This is a dated informational explanation, not legal or employment advice. California requirements can change quickly; confirm applicability and implementation with qualified counsel.