California Moves to Ban Employers From Reading Your Emotions and Neural Data

California legislators passed Assembly Bill 1883 on Sunday, sending to the governor's desk a measure that would prohibit employers in the state from using artificial intelligence tools to surveil an individual employee's emotional state or to collect an employee's neural data, as HR Dive reported. At Metaintro, we track workplace rules for what they change about your day rather than for the legal commentary, and the practical scope here is narrower than the headline suggests. This is not a ban on being monitored at work. It is a ban on one specific category of monitoring, and knowing the difference tells you what you can and cannot object to.

What would AB 1883 actually prohibit?

Two things. The first is the use of AI tools that surveil an individual employee's emotional state. The second is the collection of an employee's neural data, which the bill defines as information generated by measuring the activity of an employee's central or peripheral nervous system and that is not inferred from nonneural information. That definitional clause matters, because it draws a line between reading your nervous system directly and drawing conclusions from ordinary workplace data.

The neural data provision looks futuristic and is not. Consumer-grade devices that read brain activity already exist in headsets, earbuds and wearables marketed for focus and fatigue tracking, and the obvious workplace application is monitoring alertness in safety-critical roles. Writing the definition around nervous system measurement rather than around any particular device is what makes the clause durable, because it does not depend on the hardware staying the same. Legislating ahead of deployment is unusual, and it is the part of this bill most likely to be copied elsewhere.

The bill covers a broad list of AI tools, including those used to generate outputs that influence physical or virtual environments. Violations would carry a penalty of $500 each. There are carve-outs, and they are specific rather than general, covering operations related to the development of certain aircraft and products for national security, military, space or defence purposes. If the governor signs it, a bill of this kind would typically take effect on January 1 of the following year, and Monday was the deadline for the legislature to pass bills for his consideration according to CalMatters.

What would still be allowed?

Most of what people mean when they complain about workplace surveillance. AB 1883 would not prohibit employers from using workplace surveillance tools broadly, and it explicitly permits their use to ensure safety. Keystroke logging, badge data, call recording, productivity dashboards and location tracking in company vehicles are not what this bill addresses.

That distinction is the practical takeaway. If your employer monitors output, activity or attendance, this legislation changes nothing about your position, and the existing picture there is already extensive, with around 70 percent of large employers already watching workers in some form. What would change is the narrow and newer category of tools that claim to infer how you feel, or that measure your nervous system directly. Understanding that boundary saves you from raising an objection that the law does not actually support, and AI workplace surveillance quietly breaking how we work covers the broader category that remains lawful.

The safety exemption is the clause most likely to be argued over once this is in force. Fatigue detection in a driving role, distraction monitoring around heavy machinery and alertness checks on a night shift all sit plausibly inside it, and all of them involve inferring an internal state. An employer wanting to keep such a tool will characterise it as safety, and whether that holds will depend on whether the output is used to protect the worker or to assess them. Workers in safety-critical roles should expect that argument rather than assume the ban reaches them.

Where is emotion AI already being used?

In more places than most workers realise, and usually without announcement. Korn Ferry executives wrote in a June article that emotion AI is appearing in call centres, lobbies and offices, promoted as helpful for hiring, training and safety monitoring, and cited a projection that the market for the technology could reach $9 billion by 2030. That is a large enough number to guarantee continued deployment regardless of any single state's rules, and Korn Ferry's published research is worth reading for how the vendor case is being made to employers.

The framing in that sales pitch is worth noticing, because it is almost never surveillance. These tools are sold as coaching, wellbeing monitoring, engagement measurement or quality assurance, and the emotional inference is a feature inside a product bought for another reason. That is why workers often do not know it is running and why an employer can honestly say it has not purchased emotion AI while operating three tools that infer emotional state. Any meaningful conversation about this has to start from what the software does rather than from what it is called.

The most public example came earlier this year, when Burger King announced a rollout of an AI assistant integrated into employee headsets that can identify when workers use phrases associated with friendliness, as the Associated Press reported. The company said it intended the software as a coaching tool rather than a tracker of individual employees, which is precisely the distinction AB 1883 would make legally consequential in California. The same debate has already played out elsewhere, including when Meta stopped tracking workers to train AI after a privacy backlash.

Does emotion AI even work?

That is the part of the story most coverage skips, and it matters more than the legal question for anyone whose review might depend on it. In a March 2025 analysis, an associate professor of information at the University of Michigan wrote that emotion AI's capacity to measure emotions remains controversial and contested, adding that some research suggests adults exposed to the technology express concerns about loss of privacy and autonomy along with psychological harm.

The practical implication is uncomfortable. If a system is unreliable and is also used as an input to decisions about coaching, promotion or discipline, the errors are not evenly distributed. Tone, accent, cultural expression and neurodivergence all affect how a person presents emotionally, and a tool trained on one set of norms will misread people outside them more often. That is a fairness problem before it is a privacy problem, and it is the same underlying concern behind Colorado becoming the first state to let workers sue over AI pay decisions. Where an inaccurate inference feeds an employment decision that disadvantages a protected group, the exposure runs through existing discrimination law as well as any new statute, which is the ground the Equal Employment Opportunity Commission already covers.

How does this compare with rules elsewhere?

California would be following rather than leading. The European Union's Artificial Intelligence Act already prohibits workplace emotional monitoring, and took effect in August 2024. So the emerging position across two of the world's larger regulatory blocs is converging on the same line, which is that inferring emotion is treated differently from measuring output.

California is also not stopping at this one bill. Legislators are pursuing several AI employment measures in 2026, including SB 951, which would require employers to issue notices of mass layoffs, relocations or terminations caused substantially or wholly by AI adoption, and SB 947, which would prohibit employers from relying solely on automated decision systems to make disciplinary or termination decisions. State regulators have separately already issued rules limiting AI use in some employment decisions. Taken together that is a body of law forming around AI at work, and what the EEOC's enforcement shift means for your rights covers the federal picture running alongside it. The state's own privacy enforcement guidance is the other half of the framework, since data collected about a worker is subject to consumer privacy rules that already exist independently of any AI-specific bill.

The direction of travel across those three bills is consistent and worth naming. Each one targets a decision, not a technology. AB 1883 restricts inferring emotion, SB 947 restricts automated discipline and termination, and SB 951 forces disclosure when AI drives job losses. The common principle is that a machine may inform an employment decision but a person has to own it, which is a more workable standard than trying to regulate the software itself as it changes.

What does this mean for your career?

If you work in California, the useful move is to find out what is actually deployed on you rather than to wait for the law. Ask your employer directly what monitoring software runs on your work devices and accounts and what the data is used for, because that question is reasonable, and an employer unwilling to answer it has told you something. If a tool claims to assess your mood, engagement or attitude and that assessment feeds into a review, that is the category this bill targets and is worth raising specifically once any law is in effect. Ask in writing rather than in a conversation, keep the reply, and if the answer is that a score exists, ask to see your own. A worker who has requested their data in writing is in a materially stronger position than one who suspects a system exists, whatever the eventual legal position turns out to be.

If you work outside California, treat this as a preview rather than as irrelevant. Large employers rarely build separate systems for one state, so a California rule frequently becomes the internal default across a national workforce, in the same way that state pay transparency rules ended up reshaping job adverts far beyond their borders. "In combination with layoffs over the recent years plus the rise of required AI skills, experience is no longer enough," Lacey Kaelani, chief executive of Metaintro, told Fortune. The corollary in a monitored workplace is that documented, verifiable output is also your best protection, because it is the evidence that does not depend on how a system read your tone.

What should employers do before this takes effect?

Audit what is already running, because most organisations do not have a complete inventory. Emotion-inference features are often bundled inside call-centre quality tools, meeting software and coaching platforms rather than purchased as emotion AI, so the exposure usually sits in a product nobody thinks of as surveillance. The first task is a list of every tool that produces a sentiment, engagement or emotional score about an identifiable person.

Then separate safety uses from performance uses, since the bill's permission for safety monitoring is the meaningful carve-out and the distinction needs to be documented rather than asserted after the fact. A $500 per-violation penalty is modest individually and considerable across a workforce counted in thousands. Beyond compliance, there is a retention argument that tends to get overlooked, which is that the research on worker reactions points to autonomy loss and psychological harm, and those show up as attrition long before they show up as a claim. The California Civil Rights Department is the relevant state body for the employment discrimination questions that overlap with this, and guidance from SHRM is a reasonable starting point for building the policy itself.

Related Articles

• Your Boss's New AI Knows When You're Anxious, Inside Emotion AI's Quiet Push
• When Your Boss Uses AI to Watch How You Use AI
• Companies Now Track Employees' AI Usage in Performance Reviews
• AI Monitoring Software Flags Short Breaks, a Step Too Far?
• Inside JPMorgan's Tech-Powered Limit on Junior Banker Hours
• Can Your Boss Legally Dock Your Pay When You Show Up Late
• What Workers Should Know About Unsafe Conditions and Their Rights
• Pay Transparency Laws and What You Can Ask
• UK's New Fair Work Agency Launches, What Workers Should Know
• How to Use AI at Work Without Crossing Ethical Lines

People Also Asked

Q: What does California's AB 1883 ban?

A: It would prohibit employers from using AI tools that surveil an individual employee's emotional state, and from collecting an employee's neural data, defined as information from measuring central or peripheral nervous system activity that is not inferred from nonneural sources. Violations would carry a $500 penalty each, with carve-outs for certain aircraft and national security, military, space and defence work.

Q: Does this ban workplace monitoring in general?

A: No. The bill explicitly does not prohibit employers from using workplace surveillance tools broadly, and it permits their use to ensure safety. Activity tracking, productivity dashboards and similar tools are not what this measure addresses. The restriction is limited to inferring an individual's emotional state and to collecting neural data.

Q: When would the law take effect?

A: A bill signed by the governor would typically take effect on January 1 of the following year. The measure passed the legislature on a Sunday ahead of the Monday deadline for sending bills to the governor, so signature or veto is the remaining step before any effective date applies.

Knowing which monitoring is lawful and which is not is worth more than a general unease about being watched, and the answer increasingly depends on which state you work in. Metaintro indexes roles across states and employers so you can weigh workplace practices alongside pay when you choose where to go next. Create your free profile and compare the employers actually hiring in your field.