Estimated read time: 7 minutes
California just did something that every small business owner with an HR tool, an applicant tracker, or a productivity dashboard should pay attention to. Governor Newsom signed SB 947, widely nicknamed the No Robo Bosses Act, which bars employers from using automated decision systems as the sole basis for firing or disciplining a worker. The law does not take effect until July 1, 2027, so nobody is in trouble tomorrow. But the part that matters for you is that it applies to every employer in the state, not just the big ones with an HR department and a legal team.
If you are reading this from Texas or Vermont and thinking this is somebody else’s problem, hold that thought. California laws have a habit of becoming everybody’s laws, because it is easier to build one compliant process than fifty different ones. Here is what the law reportedly does, what it means for a business with five or fifty people, and a practical checklist you can finish in an afternoon.
Table of Contents
What SB 947 actually does
According to coverage of the signing, the law draws a line around one specific thing: letting software make or drive the call on firing or discipline. Employers cannot rely on an automated decision system as the only basis for terminating or disciplining someone. They also cannot treat the tool as the principal driver of the decision without a human check behind it.
When AI is the main input, the human reviewer has to corroborate the decision using ordinary evidence. Reports describe that as managerial evaluations, peer reviews, and personnel files. In plain English: a score from a dashboard cannot be the whole case. A real person has to look at the real record and agree, or disagree, based on something other than the machine’s output.
The law also adds transparency. If someone is fired or disciplined under a process where AI played the primary role, they are entitled to written notice that AI was involved, a description of the employee data the system analyzed, and a human contact who can explain the decision. That last one is the part most small teams will trip on, because it means someone has to actually understand what your tools are doing.
Why a small business should care
The reflex among owners of small companies is to assume that rules about algorithmic management are aimed at warehouses with productivity scanners and call centers that rank agents by the second. That is where the headlines come from. But the coverage says the law reaches all California businesses, and the definition of an automated decision system is broad enough to catch tools you may not think of as AI at all.
Think about what is quietly running in a typical ten-person company in 2026. A time tracking app that flags employees who log fewer hours than a threshold. A sales dashboard that ranks reps and highlights the bottom performer in red. A customer support platform that scores agent responses automatically. A scheduling tool that penalizes people for swaps. A new AI notetaker that summarizes one-on-ones and produces a sentiment readout. None of these were bought to fire anyone. All of them can end up as the thing a stressed manager points to when a termination conversation starts.
That is the exposure. Not that you are plotting robot firings, but that a score in a tool becomes the paper trail, and the paper trail becomes the decision.
We covered the broader hiring side of this in our AI small business hiring guide, and the compliance patchwork keeps growing, as we saw with the Colorado AI law. SB 947 is the next piece, and it lands on the discipline and termination end of the employee lifecycle.
The timeline: what happens when
The operative date reported for the law is July 1, 2027. That gives employers roughly nine months from today. It is tempting to file this under later, and for most owners that is a reasonable call for the legal deadline. The smarter move is to treat the next quarter as your quiet window to get your process in order, because the cost of doing it now is low and the cost of untangling it under pressure is high.
One caution. Coverage we reviewed did not spell out enforcement mechanisms or penalties, and details like that can matter a lot. Before you build anything permanent, read the enrolled bill text or ask an employment attorney how it will be enforced and how it interacts with existing California employment law. We are a business publication, not a law firm, and nothing here is legal advice.
What counts as an automated decision system
This is the question that decides how much work you have. Based on how the law is being described, the safest reading is a wide one. If a tool processes employee data and produces a score, ranking, flag, recommendation, or prediction that a manager could use when deciding whether to discipline or terminate someone, treat it as in scope until a lawyer tells you otherwise.
A quick sorting exercise helps. Tools that are probably in the zone include productivity and activity monitoring software, performance scoring in HR platforms, AI-generated call or meeting evaluations, attendance and shift compliance systems with automatic flags, and any custom prompt a manager runs through a chatbot to evaluate an employee. Tools that are probably outside it include a plain payroll system, a calendar, and a shared document folder, since they record facts without judging anyone.
The gray area is the chatbot problem. If a manager pastes performance notes into a general AI assistant and asks whether someone should be put on a plan, that is exactly the sort of informal use that creates risk. Nobody approved it, nobody logged it, and the output may be the main thing in the manager’s head.
A practical checklist for the next 30 days
You do not need a compliance department for this. You need a short list and an owner. Here is a version that fits a small team.
- Inventory your tools. Write down every system that touches employee performance or attendance data. Include the ones a single manager adopted on their own. This list is the foundation for everything else.
- Mark what scores or ranks people. For each tool, note whether it produces anything that looks like a rating, flag, or recommendation about an individual.
- Set a human review rule. Write one sentence of policy: no termination or discipline decision may rest on a tool’s output alone. A named person reviews the underlying record, such as past evaluations, peer feedback, and the personnel file.
- Document the review. Keep a short note for every discipline or termination showing what the human looked at and why they agreed. If the decision is ever questioned, that note is your best friend.
- Pick the explainer. Decide now who the human contact is when a worker asks how the decision was made. That person needs to understand what each tool does and what data it uses.
- Ask vendors two questions. What employee data does this tool analyze, and can you give me a plain-language description of how it produces its outputs? Vendors selling into California should have an answer ready.
- Update your handbook and AI policy. If you have not written an AI policy yet, our AI use policy template is a good starting point, and you can add a short section on employment decisions.
Notice what is not on the list: ripping out your software. The law, as described, is about how you use the tools, not whether you own them. A dashboard that helps you spot a problem early is fine. A dashboard that is the whole reason someone lost their job is the thing California is trying to stop.
The bigger picture for owners everywhere
There is a pattern worth naming. Colorado went first on broad AI accountability. California is now targeting a specific high-stakes moment, the end of someone’s job. Other states are watching, and employers who operate across state lines or hire remote workers in California will feel it even if their headquarters is somewhere else.
The upside is that the compliant behavior is also just good management. Decisions about people should rest on more than a number. Employees who get a real explanation are less likely to feel blindsided and less likely to dispute. A short documented review protects you whether the rules change again or not.
The downside is that the compliance load keeps landing on the businesses least equipped to absorb it. A five-person agency does not have a general counsel. That is why a lightweight written process, done early, is worth more than a perfect one done in a panic in June 2027.
If you are also thinking about how AI is changing your hiring, this is a good moment to read our guide on how to hire your first employee, since the basics of documented, human-led people decisions apply from day one.
The bottom line
SB 947 does not ban AI at work. It bans letting AI be the only reason a person loses their job, and it asks employers to be able to explain what happened. For a small business, the work is mostly a short inventory, a one-line policy, a review habit, and a named human who can answer questions.
Nine months sounds like forever until it is not. Spend an afternoon on the checklist above this month, talk to an employment attorney about the enforcement details, and you will be ahead of most of your competitors on a rule that is likely to spread.
Frequently Asked Questions
What is California SB 947?
SB 947, often called the No Robo Bosses Act, is a California law that bars employers from relying solely on automated decision systems to fire or discipline workers. Reports say it requires human review that corroborates AI-driven decisions and takes effect July 1, 2027.
Does SB 947 apply to small businesses?
Coverage of the law says it applies to all California employers. We did not see a small business exemption described, so assume it covers you if you employ workers in California and confirm with an employment attorney.
When does the law take effect?
The operative date reported is July 1, 2027, which gives employers about nine months to review their tools and processes.
What do employers have to tell workers?
Reports describe a notice requirement for people fired or disciplined under a process where AI played the primary role: written notice that AI was involved, a description of the employee data analyzed, and a human contact who can explain the decision.
Can I still use AI tools for HR?
Yes. The law is described as targeting AI-only or AI-driven firing and discipline decisions, not the use of software generally. The key is that a human reviews the underlying evidence and the decision is documented.
Is this legal advice?
No. This article is general business information based on published coverage. Read the bill text and consult an employment attorney about how SB 947 applies to your company.
Related Coverage
- AI Small Business Hiring Guide: how AI is reshaping who you hire and how
- AI Use Policy for Employees: a template you can adapt to cover employment decisions
- Colorado AI Law and Small Business Compliance: the other state law shaping how you use AI on people
Faceted Media Magazine covers business, AI, and entrepreneurship for the people building what’s next.