California bans employers from using AI tools to collect workers' neural data

By Inside BCI · October 2, 2026

An AI-enabled headband that tells a supervisor when a warehouse picker’s attention is drifting, or call-centre software that scores an agent’s voice for frustration, becomes off-limits for most California employers on 1 January 2027 unless it falls within the law’s safety exception. Governor Gavin Newsom signed AB 1883 on 30 September, the last day he could act on bills this session.

What Newsom signed

AB 1883, by Assemblymember Isaac Bryan (D-Los Angeles), adds a new part on workplace surveillance tools to the California Labor Code. It was chaptered as Chapter 853, Statutes of 2026. The prohibition has two prongs. An employer may not use a workplace surveillance tool that uses artificial intelligence on employees to recognise, infer or predict an individual’s emotional state, or to collect neural data.

Newsom signed it alongside AB 1331, which restricts workplace surveillance tools in workplace restrooms, and SB 947, which from 1 July 2027 stops employers relying solely on automated decision systems to discipline or fire workers. The governor’s office billed the package as first-in-the-nation worker protections.

In a signing message covering AB 1331 and AB 1883, Newsom wrote: “Bad-actor employers can now collect vast amounts of granular, personal information, leaving employees vulnerable to manipulative or abusive practices.” He encouraged the Legislature “to refine these two measures to address ambiguities and possible unintended consequences,” and said AB 1883 “lacks key definitions, which may create confusion about which tools are covered.”

What the law covers

The law defines neural data as “information that is generated by measuring the activity of an employee’s central or peripheral nervous system, and that is not inferred from nonneural information.” The wording tracks the definition California wrote into its consumer privacy law through SB 1223, which made neural data sensitive personal information from 1 January 2025. EEG headbands, caps and earbuds fall inside it. Wearables that read muscle or skin signals are harder cases, since the definition excludes information inferred from nonneural sources.

A workplace surveillance tool is any “system, application, instrument, or device that collects or facilitates the collection of employee data, activities, communications, actions, biometrics, or behaviors by means other than direct observation by a person.” The listed examples, from video and audio surveillance to continuous incremental time-tracking tools and geolocation, are an open list. Because the definition includes communications, software that infers mood from email, chat or call audio could fall under the emotion prong.

The ban applies only where the tool uses artificial intelligence, defined as an engineered or machine-based system that can, for explicit or implicit objectives, infer from the input it receives how to generate outputs that can influence physical or virtual environments. Many commercial focus-scoring and sentiment products are likely to meet that test. The neural-data prong is triggered by collection, so an employer does not have to use the data in a decision to breach it.

The law covers private employers and public ones, including all branches of state government, the University of California, the California State University, cities and counties. It protects employees. Senate amendments in August removed independent contractors, who were covered in earlier drafts.

Two exceptions apply. The law does not prohibit an employer from using a workplace surveillance tool “to ensure safety”, and it does not reach operations where a tool is reasonably necessary to comply with a federal statute, federal regulation or binding federal contract covering aircraft for the national airspace or national security, military, space or defence work. “Ensure safety” is not defined. A broad reading could protect fatigue monitors on truck drivers and heavy-equipment operators even where they collect neural data. A narrow reading could leave those deployments exposed.

Enforcement sits with the Labor Commissioner, who can investigate, order temporary relief, issue citations and sue, and with public prosecutors. In those actions the enforcer can seek injunctive relief, including punitive damages, and attorney’s fees. Penalties run up to $500 per violation, with no rule on whether a violation is counted per employee, per day or per tool. The Senate struck the private right of action that earlier drafts gave workers. The provision allowing “an employee” to recover a penalty leaves room for lawyers to argue for claims under California’s Private Attorneys General Act, a theory no court has tested.

Where California now draws the line on neural data

On 27 September Newsom vetoed AB 1542, which would have banned businesses from selling or sharing sensitive personal information, a category that has included neural data since January 2025. Consumers keep an opt-out over the sale of their neural data. The same day he signed SB 354, which from 1 July 2028 limits how insurers may process or share sensitive personal information, including neural data. On 30 September he signed AB 1883.

The pattern suggests Sacramento is prepared to set firmer rules where a person cannot easily refuse a device or a data use, as with an employer or an insurer, while leaving consumers to manage their own data through opt-outs.

How it compares with the EU ban

The EU AI Act has prohibited AI systems that infer emotions in the workplace since 2 February 2025, with exceptions for medical or safety reasons. The European Commission’s guidelines confine that ban to emotion recognition based on biometric data and exclude physical states such as fatigue. AB 1883’s emotion prong carries no biometric condition, its neural-data prong covers AI-enabled EEG tools whatever they measure, and it has a safety exception with no medical one. For multinationals, a single policy that switches emotion inference off at work could prove simpler than separate California and EU configurations.

Where this sits on the four-jurisdiction posture map

Issue 01 of the Dargentic Intelligence Report classified four global regulatory postures on brain-computer interfaces: Rights-First (Chile), Horizontal Regulator (EU with AI Act and GDPR), State-Patchwork (US), and Industrial Builder (China and South Korea). Inside the US column, Colorado, California, Montana and Connecticut have enforceable neural-data laws. Vermont’s Act 101 (H.814), in force since 18 May 2026, recognises neural rights without attaching obligations or enforcement, and Vermont’s separate Act 145 (S.71), signed 16 June 2026, requires consent to process or sell neural data from 1 January 2028. Delaware adds neural data to its sensitive data category from 1 January 2027 and allows its sale only when strictly necessary.

AB 1883 adds a different kind of rule to that column: a labour law, enforced by a labour regulator, that prohibits one use of neural data, subject to the safety and federal-contract exceptions, with no consent or opt-out route for the employer. Where AI-enabled tools are involved, California now treats neural data more strictly inside the employment relationship than in the consumer market.

What it means for workplace neurotech vendors

The day after the signing, Boston-based Neurable put Neurable One on sale in the United States at $499. The over-ear headphones carry 12 EEG sensors in the ear pads, score focus and cognitive load, and store brain activity whenever they are worn with the app open, a feature Neurable calls Brain Data Banking. Neurable also licenses its sensing technology and names helmets and hearing protection among its target form factors. Separately, it offers Neurable for Teams, an employer programme built around the MW75 Neuro headphones it made with Master & Dynamic, pitched at raising productivity and reducing burnout.

A worker who buys a pair for personal use sits outside AB 1883, which governs what employers do. An employer that issues EEG headphones to California staff to lift productivity or manage burnout could be collecting neural data with an AI-enabled tool for a purpose that is not obviously safety, and from 1 January 2027 could face penalties. A fatigue-monitoring helmet or ear defender on a hazardous site is the kind of deployment the safety exception could protect, particularly where the employer can show the device serves a safety purpose.

Not disclosed

How the Labor Commissioner will read “ensure safety” is not known. How a single violation is counted for the $500 penalty is not stated in the law. How many California employers currently run AI-enabled neural-data or emotion-inference tools is not public. Whether Neurable’s Brain Data Banking stores raw EEG or only derived metrics, and whether any California employer uses Neurable for Teams, is not disclosed.

What to watch

Whether vendors of contact-centre analytics, collaboration and wellbeing platforms and fatigue wearables switch off emotion-inference and neural-data features for California employees before 1 January 2027, or recast them as safety tools.

Whether the Legislature acts on Newsom’s request to refine AB 1883 in the 2027 session, starting with the undefined safety exception.

Whether the first Labor Commissioner citation or public-prosecutor action arrives in 2027, and whether employee-side lawyers test the penalty provision through the Private Attorneys General Act.

Whether other states copy AB 1883’s two-prong text.

Sources

Primary:

Coverage:

Cross-reference to prior InsideBCI coverage:

policy California AB-1883 Isaac-Bryan Gavin-Newsom neural-data emotion-recognition workplace-neurotech workplace-surveillance Labor-Commissioner AB-1331 SB-947 AB-1542 SB-354 SB-1223 CCPA PAGA EU-AI-Act Neurable Neurable-One State-Patchwork-posture Issue-01