California bars employers from letting AI alone fire or discipline workers

California workers can no longer be fired by an algorithm with nobody accountable behind it. Governor Gavin Newsom has signed SB 947, known as the No Robo Bosses Act, which stops employers from relying only on automated systems to fire or discipline workers. The office of the bill's author, State Senator Jerry McNerney, called it the first law of its kind in the country.

The law does not ban AI from the workplace. Employers can still use software to track performance, flag problems or rank workers. What changes is the final step: when a machine's output drives a decision to punish or dismiss someone, a person has to check it, sign off on it and tell the worker what happened.

What the No Robo Bosses Act covers

The law is built around the idea of an automated decision system. According to The National Law Review, that means any computational process derived from machine learning, statistical modeling, data analytics or artificial intelligence that produces a simplified output, such as a score or a recommendation, used to assist or replace human decisions that materially affect people.

That definition is broad on purpose. It reaches well beyond chatbots. Productivity scores from monitoring software, flags from a scheduling system that tracks lateness, or rankings from a performance analytics tool could all qualify if they feed into a disciplinary or termination decision.

Everyday tools such as spam filters, firewalls, antivirus software and calculators are excluded, the National Law Review reported, along with databases and datasets.

What employers must do

The National Law Review reported that when an automated decision system plays a primary role in a firing or disciplinary decision, a human must review it. That person has to check the system's output against other evidence, such as manager evaluations, personnel records, work product, peer reviews and witness interviews. If the output cannot be verified, it cannot be used. The same applies if the reviewer finds the output inaccurate, incomplete or misleading.

In practice, this means a manager cannot simply forward a system alert to human resources and call it a decision. Suppose a warehouse system flags a worker for low pick rates. Under the new rules, someone would need to look at whether the numbers hold up: Was the worker reassigned that week? Was equipment down? Do supervisor notes back up the flag? If the answer is unclear, the flag cannot be the basis for discipline.

The written notice

Employers must also give the worker written notice that an automated system was involved. According to the National Law Review, the notice has to:

• Explain that the employer relied on an automated system.
• Confirm that a human reviewed the decision.
• Give a contact for questions.
• Tell the worker they can request a plain description of the data used about them.

The National Law Review also reported that the notice must state that retaliation is prohibited.

Uses that are off limits

The law also bars certain uses outright, according to the National Law Review. Employers cannot use these systems to infer a worker's protected status under California's Fair Employment and Housing Act, to predict that a worker will exercise legal rights and take adverse action on that basis, or to get around employment, labor, safety or civil rights laws.

When it starts and who enforces it

The law takes effect on July 1, 2027, the National Law Review reported. The California Labor Commissioner and public prosecutors can enforce it.

There is one notable carve out. The National Law Review reported that qualifying collective bargaining agreements can waive the provisions if they include their own protections around algorithmic management. That gives unions room to negotiate terms that fit their industries, as long as workers end up with comparable safeguards.

What supporters said

"No worker should ever be fired or disciplined by a robo boss," McNerney said in a statement. "AI must remain a tool controlled by humans, not the other way around."

Lorena Gonzalez of the California Federation of Labor Unions said workers across the state "have demanded that our state lead the way in regulating AI in our workplaces."

How it fits the national picture

California is moving on specific, enforceable rules while federal policy is heading in a different direction. In Washington, the administration has launched a new Super Intelligence Force focused on coordination and competition, and AI companies have signed a voluntary White House safety accord with no penalties attached. A state law with an enforcement agency behind it is a very different tool.

Because so many large employers operate in California, the law may shape practices elsewhere too. A company that builds a human review and notice process for its California staff may find it simpler to use the same process everywhere. We have argued that workers deserve to know when software is judging them, and this law turns that principle into a legal requirement for one large state.

What employers should do now

July 2027 sounds far off, but mapping how software touches employment decisions takes time. A practical starting list:

1. Inventory your tools. List every system that scores, ranks, monitors or flags workers, including features inside larger HR or scheduling platforms.
2. Trace the decisions. For each tool, note whether its output has ever fed into discipline or termination, and who acted on it.
3. Name reviewers. Decide who will corroborate automated outputs, and make sure they have access to the underlying records.
4. Draft the notice. Prepare a template that covers each required element, including the contact person and the data request process.
5. Ask your vendors. Find out what data their systems use and whether they can produce a plain description of it on request.
6. Check union agreements. If you have a collective bargaining agreement, review whether it addresses algorithmic management.

What workers should know

If you work in California, you will be entitled to know when software helped decide your discipline or dismissal, and a person will have to stand behind that decision. Once the law takes effect, if you receive a notice, you can ask for a description of the data used about you. Keep copies of any notices and your own records of your work, since those can matter if a decision is disputed. For broader context, see our AI coverage.

Frequently asked questions

Does the No Robo Bosses Act ban AI in hiring?

The provisions described here focus on discipline and termination decisions. They require human review and written notice when an automated system plays a primary role in those decisions.

When does the law take effect?

July 1, 2027, according to The National Law Review.

Who enforces it?

The California Labor Commissioner and public prosecutors.

What counts as an automated decision system?

Any computational process based on machine learning, statistical modeling, data analytics or AI that produces a simplified output used to assist or replace human decisions with a material impact on people. Spam filters, firewalls, antivirus software, calculators, databases and datasets are excluded.