Four New California Laws Reshape the Rules for Workplace AI

As one of the nation’s leaders in artificial intelligence (AI) and technology regulation, California is implementing guardrails to protect consumers and employees while increasing transparency in the use of AI.

Among the many AI-related regulatory bills approved this session, the state legislature passed four measures designed to protect workers from AI-driven decision-making and surveillance. Governor Gavin Newsom signed them into law on September 30, 2026. Three of the four laws will take effect on January 1, 2027, with the fourth becoming effective on July 1, 2027.

For employers with California employees, the message is straightforward: algorithmic management tools, workplace monitoring technology, and AI-driven workforce reductions now carry specific compliance obligations, real enforcement exposure, and documentation demands that most human resources (HR) departments have not yet built into their processes. The sooner employers start assessing their workplace and identifying areas for compliance, the better positioned they will be.

What You Need to Know

• Guardrails on Automated Decision-Making: Employers may not rely solely on an automated decision-making (ADS) tool’s output to discipline or terminate employees. When an employer relies primarily on an ADS output for such a decision, the employer must ensure that a human corroborates the ADS output, notify affected employees after the fact, and, upon request, provide employees a description of their own data used by the tool.
• Restrictions on Workplace Surveillance: The new laws restrict employers from using “workplace surveillance tools” to monitor employees in workplace bathrooms and prohibit the use of AI-driven workplace monitoring that predicts or infers employees’ emotions or gathers their neural data.
• Mandatory Disclosure of Technology Displacements: Amendments to the California Worker Adjustment and Retraining Notification Act (Cal/WARN) will require employers to notify individuals and certain agencies when a mass layoff, relocation, or closure occurs because of the adoption of an AI system or automation that replaces workers.
• Meaningful Enforcement Exposure: The new laws carry civil penalties of $500 per violation, provide for enforcement actions by the Labor Commissioner or public prosecutors, and include the potential for punitive damages and attorneys’ fees. AB 1883 also gives employees a private right of action. Under SB 947, employers that cannot document their human review of an adverse employment decision may find themselves at a distinct disadvantage in a dispute.

(1) SB 947: The No Robo Bosses Act of 2026

As discussed in our previous Insight, SB 947—the “No Robo Bosses Act”—will restrict how California employers can use ADSs. This was the California Legislature’s second attempt at this issue, as Governor Newsom vetoed a similar bill, SB 7, in October 2025, citing overly broad notification requirements. The 2026 version drops the broad advance-notice obligations that drew that criticism. The 2026 law is narrower but still consequential. Employers have until the law’s July 1, 2027, effective date to ensure that their use of an ADS complies with several key requirements.

Under the No Robo Bosses Act, an ADS includes any machine-learning, statistical modeling, data-analytic, or AI-driven process used to assist or replace discretionary decision-making. Employers must recognize the functional application of this definition: what matters is how it uses a tool, not how a vendor markets it. A tool purchased for one purpose (such as scheduling or security) becomes an ADS for purposes of the No Robo Bosses Act when its scores, flags, or uses recommendations to inform discipline or termination. The No Robo Bosses Act excludes basic technologies, such as spam filters, firewalls, antivirus software, and databases. Attendance and scheduling trackers, productivity and activity monitoring software, algorithmic rating and ranking tools based on work output, customer ratings, or quotas, and security monitors potentially constitute ADSs. 

Prohibitions and New Obligations

Under the No Robo Bosses Act, employers may not use an ADS to:

• Make disciplinary or termination decisions based solely on ADS output. The new law, adding Sections 1520 et seq. to the California Labor Code, plainly states that employers “shall not rely solely on an ADS when making a disciplinary or termination decision” (emphasis added).

Despite this prohibition, the new Section 1522 contemplates that employers will continue to rely on an ADS and may rely “primarily” upon an ADS output to make decisions. (Although the new law references the concept of an employer “primarily” using or relying “upon an ADS” six times, it does not define the term.)

Employers that “primarily” rely on ADS data to make an adverse employment decision will need to assign a human reviewer to corroborate that decision by checking either the original data underlying the ADS output or other relevant corroborating or supporting information, such as:

• Supervisory or managerial evaluations
• Personnel files
• Employee work product
• Peer reviews
• Witness interviews (including relevant online customer reviews)

Practically, a manager who solely reads a dashboard score and signs off has not corroborated anything. The human reviewer should independently compare the ADS output against at least one other source and have authority to reject the dashboard score. If the human reviewer cannot corroborate the output, or concludes that it is inaccurate, incomplete, or misleading, the employer should not use it as the basis for the decision.

• Prevent compliance with any labor and civil rights laws. The statute broadly incorporates federal, state, and local laws and regulations governing labor, occupational health and safety, employment, and civil rights. An employer’s reliance on any ADSs that interfere with compliance may result in a violation of the No Robo Bosses Act.
• Infer protected status. Using an ADS to conclude that an employee has protected status under the California Fair Employment and Housing Act violates the law—because employers could use that information to discriminate against employees with certain protected status. Employers increasingly use ADSs to make employment decisions, so California has created guardrails against the biases that ADSs can cause.
• Retaliate through automation. Using an ADS to anticipate an employee’s exercise of legal rights and then taking adverse action is unlawful under the No Robo Bosses Act.

The No Robo Bosses Act also imposes several new obligations on employers who use an ADS:

Post-Use Notices: Employers must provide employees with written notices when their disciplinary or termination decisions are based “primarily” on an ADS. These notices must:

• disclose that the employer primarily relied on an ADS to reach the disciplinary or termination decision;
• confirm that a human reviewed and corroborated the ADS output;
• provide contact information for a person who can supply more information about the decision; and
• include information about the employee’s rights, including the right to access their data (as explained below) and the legal prohibition on retaliation.

The law states that compliance with these notice requirements will exempt employers from substantially similar notice obligations imposed by other California laws. Notably, the No Robo Bosses Act does not impose a general posting requirement.

As of July 1, 2027, employers must have in place the template, the delivery method, and the designated contact person before the first covered decision. As a best practice, we recommend written notices that use plain language and appear in the language the employer ordinarily uses with the employee. Employers should deliver the notices separately from the disciplinary or termination letter.

Data Access Rights: When an employer has “primarily” used an ADS to make an adverse decision, employees have a right to “a meaningful, objective description” of their own data used by the ADS. The law requires employers to safeguard privacy by anonymizing personal information when providing such a description of “employee data,” which is defined as “any information that identifies, relates to, or describes an employee, regardless of how the information is collected, inferred, or obtained.”

The law does not address questions of timing, such as how long an employer should retain data or the duration of an employee’s right to obtain a description of their data after a decision. It is also vague as to whether employees’ right to request their own data applies only when the employee was the subject of an adverse action or any time the employer “primarily” relied on an ADS to make a disciplinary or termination decision. Pending guidance from the Labor Commissioner, we recommend that employers adopt a conservative retention period for ADS inputs, outputs, and review records tied to discipline and termination decisions (at a minimum, long enough to respond to a request and defend a claim), and decide now who will respond to data requests and how coworker and customer information will be anonymized.

Other Features of the No Robo Bosses Act

The No Robo Bosses Act applies to private and public-sector employers, and it reaches labor contractors, so a staffing arrangement will not necessarily insulate a business from responsibility for tools it uses. A collective bargaining agreement may waive the No Robo Bosses Act’s requirements only in clear and unambiguous terms and only where the agreement provides protection from algorithmic management. The No Robo Bosses Act also contains a carve-out for ADS use mandated by certain federal aviation, national security, military, space, and defense requirements, and it does not displace more protective local laws. Importantly, the No Robo Bosses Act shifts the burden of proof: once an employee shows that an employer used an ADS, the employer must demonstrate that it did not “primarily" rely on the ADS or that it complied with the human review and notice requirements.

Relationship with the CCPA – Dual Compliance

Employers already subject to the California Consumer Privacy Act (CCPA) must comply with both frameworks:

• The CCPA requires pre-use notices before using automated decision-making technology (ADMT) for significant employment decisions. ADMT has a broader definition than what is considered ADS. Please see our previous blog post for more information on ADMT and the pre-use notice requirements.
• The No Robo Bosses Act requires post-use notices regarding ADS-influenced decisions.
• The No Robo Bosses Act and the CCPA have overlapping but complementary notice requirements. Employers subject to the CCPA must still comply with the CCPA’s privacy-related ADMT regulations even when complying with the No Robo Bosses Act’s notice requirements. In practice, employers should design a single, integrated ADS governance process (covering inventory, pre-use notice, human review, post-use notice, and data-access requests) rather than managing the two regimes in silos.

The timeline for compliance with each law is different:

• CCPA’s Pre-Use Notice: Effective January 1, 2027
• No Robo Bosses Act’s Post-Use Notice: Effective July 1, 2027

The No Robo Bosses Act is the first law in the nation to establish mandatory human oversight in employment decisions driven by an ADS. Because other states are likely to follow California’s lead, multistate employers should consider whether to apply the No Robo Bosses Act’s human-review protocol uniformly rather than only in California.

(2) AB 1331: No Surveillance in Bathrooms

AB 1331 prohibits California employers from using workplace surveillance tools to monitor bathrooms, except when required by court order. The law encompasses video, audio, time-tracking, geolocation, and electronic monitoring devices, including those with embedded AI. The law does not extend to other workplace facilities, such as cafeterias, locker rooms, or break areas.

Employees retain the right to leave surveillance tools behind when entering bathrooms; if, however, an employee voluntarily chooses to take a surveillance tool into a bathroom, that does not cause a violation.

An employer may maintain a policy that requires employees to always possess workplace surveillance tools, including while inside a bathroom, if those tools do not detect or record video or audio and do not have AI attached or embedded.

Implementation Challenges

The broad definition of “workplace surveillance tools” raises concerns about AB 1331’s practical application to, for example:

• Employee personal devices (such as phones, etc.) brought into restroom facilities
• Key card or badge access systems requiring employees to use electronic identification to access bathrooms

In addition, because many vendors bundle multiple functions (security, HR, facilities management), disentangling surveillance from non-surveillance functions could be complex.

Governor Newsom raised such ambiguities to the California State Assembly in a signing memo (the “Signing Memo”), issued when he approved AB 1331 and AB 1883 (discussed below). The Signing Memo encourages lawmakers to refine the statutory language to clarify the law’s scope.

Practical steps: Pending the clarification the Governor has requested or a court’s interpretation of the law, employers should (i) disable or geofence bathroom entry, exit, and location tracking in badge readers, wearables, and workforce-management platforms; (ii) obtain written confirmation from vendors that bathroom areas are excluded from video, audio, and location capture; and (iii) avoid disciplining employees based on “time away from station” data that effectively reconstructs bathroom use. Employers may also continue to prohibit the use of video or other recording devices by employees in bathrooms and similar facilities.

(3) AB 1883: Restricting AI-Enabled Workplace Surveillance

Despite reservations in the Signing Memo about AB 1883, the Governor enacted the legislation, which restricts how California employers may use workplace surveillance tools that rely on AI. The law prohibits employers from using AI in conjunction with workplace surveillance tools to recognize, infer, or predict an employee’s emotional state or collect “neural data”—i.e., information generated by a human nervous system. The law, however, is narrowly tailored to these specific uses, still permitting non-AI monitoring and explicitly allowing surveillance tools for safety. Whether the safety exception is broad or narrow remains an open question. Unless or until lawmakers or a court clarifies the scope of this exception, employers should work with counsel and build their monitoring programs based on their individual situation and tolerance for legal risk. 

Under this law, an “employee” is simply “a person employed by an employer.” This definition, on its face, does not extend to applicants, interviewees, or prospective or former employees, but is not necessarily limited to employees located in California.

A “workplace surveillance tool” is any system, application, instrument, or device that collects employee data, activity, communication, biometrics, or behaviors by means other than direct observation by a person. Examples include video or audio surveillance, continuous incremental time-tracking tools, geolocation systems, electromagnetic tracking, photoelectronic tracking, and certain photo-optical devices. The Signing Memo expressed concern that this broad definition may create confusion about which tools the law covers and encouraged a rewrite.

While the law appears to apply to a wide variety of workplace technologies, including such commonplace devices as GPS trackers and camera systems, the limitations created by AB 1883 are rather specific. The law’s prohibitions on using AI to predict or infer an employee’s emotional state or to collect neural data are categorical, even for well-intentioned uses like stress monitoring and wellness programs. But the law carves out certain national security concerns, including the development of aircraft and other products and services for military, space, or defense purposes. Importantly, it also expressly permits workplace surveillance tools, including those that employers use to ensure safety, if they do not violate the prohibitions on neural data gathering and emotion prediction.

Practical points: First, the prohibition turns on what the employer uses AI to infer, not on whether a tool exists. Ordinary video conferencing or meeting recording should not, by itself, run afoul of the law, but layering “sentiment,” “engagement,” or “stress” analytics onto an employee’s voice, face, or biometrics would. Call-center sentiment analysis, driver-monitoring systems that score fatigue or mood, and wellness wearables are the likeliest trouble spots. Second, because the statute defines “employee” without limiting it to California, employers with remote or out-of-state workers should not assume that the law stops at the border; a conservative approach is to apply the prohibition to all workers of a California employer pending further guidance. Third, the statute does not, on its face, reach applicants, so AB 1883 does not squarely cover AI interview tools that analyze a candidate’s expressions or tone. Such tools, however, remain subject to anti-discrimination laws, the CCPA’s ADMT regulations, and a growing number of state and local AI hiring laws.

(4) SB 951: The California Employee Technological Displacement Act

SB 951 amends Cal/WARN to impose additional notice requirements when an AI system or other automated technology that replaces or automates jobs causes a mass layoff, relocation, or termination, in whole or in substantial part. The bill makes California the second state in the nation to enact legislation requiring employers to disclose the displacement of workers by automated technology or AI, following Connecticut.[1]

Cal/WARN applies to employers that operate industrial or commercial facilities with 75 or more employees at any time in the 12 months preceding a layoff. Under the existing law, covered employers must provide 60 days’ written advance notice before a mass layoff, relocation, or termination takes effect. As amended by SB 951, Cal/WARN will require employers to include AI-specific disclosures in their Cal/WARN notices when AI or automated technology causes a mass layoff, relocation, or termination “in whole or in substantial part.”

Under such circumstances, Cal/WARN notices will need to include:

• the statement “This notice is for a technology displacement” at the top of the notice;
• the number, classification, occupation, and work location of employees whose positions will be automated;
• a description of the specific job functions that AI or other technology will automate; and
• the specific category or type of AI system or automated technology causing the displacement.

The bill also adds a temporary provision requiring California’s Employment Development Department to submit a report to the Legislature on AI’s effects on business hiring practices at the state and regional level by January 1, 2028. The report must include recommendations and/or analysis regarding public notice requirements for employers ending hiring due to AI, hiring trends across industries and occupations based on AI exposure, how businesses use AI and its effects on hiring practices, and existing research on the effects of AI on businesses’ hiring needs and practices. This reporting requirement expires January 1, 2029.

Practical implications: Any notice that labels a layoff a “technology displacement” and identifies the functions automated and the category of AI involved is, in effect, a public statement about causation, and it could be read as the cause of the layoff by affected employees, their counsel, unions, regulators, and the press. Employers planning a reduction in force should (i) decide early whether AI or automation is a cause “in whole or in substantial part”; (ii) document the business rationale and selection criteria; (iii) align the Cal/WARN notice with separation agreements, Older Workers Benefit Protection Act disclosures for employees age 40 and older, and external communications; and (iv) coordinate notices across jurisdictions, particularly because other states’ WARN laws are also beginning to address AI-driven layoffs.

New Provisions for Enforcement and Remedies

The first three of these laws add new provisions to the California Labor Code addressing enforcement. The No Robo Bosses Act, the restroom surveillance ban, and the neural data collection ban all vest the Labor Commissioner and public prosecutors with enforcement powers, create civil penalties of $500 per violation, and provide for temporary or preliminary injunctive relief (court orders to stop the violation), punitive damages, reasonable attorneys’ fees and costs, and other remedies available under the statutes. Because the laws assess penalties per violation, exposure scales with the number of affected employees and decisions. Because of the burden shift under the No Robo Bosses Act, contemporaneous documentation of human review will be the single most important defensive measure.

The neural data collection ban, unlike the other two laws, also provides employees with a private right of action, meaning aggrieved individuals will be able to sue their employers. The law, however, limits recovery. An employer may be liable for a civil penalty or the statutory penalty, but not both, for the same violation. We anticipate that plaintiffs’ counsel will test this provision, particularly in call-center, logistics, and wellness-program settings where developers often bundle emotion-recognition features into otherwise routine tools.

What California Employers Should Do Now

With the January 1, 2027, effective date for most of this legislation only a few months away, California employers that rely on AI or other automated decision tools in their workflows must act quickly to assess the impact of this heightened regulatory environment. Although the No Robo Bosses Act does not become operative until July 1, 2027, the work it requires (inventorying tools, building a review process, and training managers) takes time, and much of it overlaps with what the CCPA’s automated decision-making rules require by January 1, 2027.

Employers should consider taking the following steps now:

• Audit AI, ADS, automation, and workplace surveillance tools that the new laws may cover. Include tools embedded in HR information systems and scheduling, timekeeping, security, video-conferencing, and call-center platforms, and ask vendors in writing whether they have any feature that scores, ranks, flags, or infers emotion or location. Where practical, conduct the audit at counsel’s direction to help preserve privilege.
• If using an ADS to make employment decisions, do the following:
  • Evaluate existing processes and determine whether decision-makers make decisions “primarily” based on an ADS output, ensuring consistency of such use. Adopt a written internal definition of “primarily” (a conservative approach treats any decision in which an ADS output was a trigger or principal factor as covered) and apply it uniformly.
  • Adjust as necessary to incorporate and document human corroboration of an ADS output prior to making final determinations.
  • Train managers and HR reviewers on what counts as meaningful corroboration.
  • Give reviewers the authority to reject an ADS output.
  • Develop and use a standard review form that records the independent sources consulted and the reviewer’s conclusion.
  • Remind decision-makers of the No Robo Bosses Act’s prohibition on retaliation.
  • Prepare model post-use notices for mandatory disclosure regarding ADS use in the decision-making process by July 1, 2027; if subject to the CCPA, separately prepare pre-use ADMT notices and related compliance steps by January 1, 2027.
  • Draft a protocol for responding to employee requests for a description of their ADS data, including anonymizing co-worker and customer information, and set retention periods for ADS inputs, outputs, and review records.

• Conduct a comprehensive audit of surveillance infrastructure to prepare for compliance with the new prohibition on bathroom monitoring and to address employees’ right to disconnect from surveillance while in restroom facilities. Be aware that features that track employees’ entry and exit of bathrooms, including timestamps or frequency reports, may create a compliance problem.
• Work with vendors to determine whether tools such as wearable devices, badges, personal alarm systems, and AI-powered identification systems create a risk of violating requirements imposed by AB 1331’s prohibition on bathroom monitoring.
• Update any policies regarding workplace badges or other monitoring devices. Revise handbooks and device policies to reflect the bathroom restriction, employees’ right to leave surveillance tools behind, and the limits on emotion interference and neural-data collection, and to document the legitimate safety, security, and operational purposes of the monitoring that remains lawful.
• Assess workplace use of neural data or emotion-monitoring technology, including seemingly benign programs such as wellness trackers, in light of AB 1883. Where emotion-inference features are bundled into an otherwise useful tool, ask the vendor to disable them for California employees and consider doing so company-wide given the statute’s ambiguous geographic reach.
• If subject to Cal/WARN, be prepared to report any layoffs or furloughs of workers displaced by automated technology or AI and include the new technology-displacement disclosures in the required notice. Update reduction-in-force checklists and templates accordingly and involve employment counsel early.
• Review vendor and technology contracts. Require vendors to disclose whether their tools generate scores, rankings, or inferences about employees’ emotions; to support your review, notice, and data-access processes; and to indemnify you for non-compliant features.
• Review collective bargaining agreements for provisions bearing on algorithmic management.
• Expect further developments, including amendments to the new statutes, rulemaking, and guidance. Be aware that the Signing Memo on AB 1331 and AB 1883 invites clean-up legislation; the Labor Commissioner may issue guidance on undefined terms such as “primarily”; and California’s privacy regulators are separately implementing the CCPA’s automated decision-making rules.

We will continue to monitor and report on any developments.

* * * *

For additional information about the issues discussed in this Insight, please contact the attorney(s) listed on this page or the Epstein Becker Green Employment, Labor & Workforce Management attorney who regularly handles your legal matters.

Staff Attorney Elizabeth A. Ledkovsky assisted with the preparation of this Insight.

ENDNOTE

[1] In 2025, New York State’s WARN reporting system added “artificial intelligence” to the list of reasons for layoffs and closures that employers may disclose.