California employers can use AI to help make employment decisions, but starting next year, they cannot let the algorithm do the firing. Governor Gavin Newsom signed Senate Bill 947 on September 30, creating what supporters have dubbed the “No Robo Bosses Act.” The new law draws a very California line in the sand: automated systems can inform discipline and termination decisions, but they cannot replace meaningful human judgment when someone’s job is at stake.

That sounds straightforward. It may not be. Many employers already use software that scores productivity, flags attendance, ranks workers, identifies possible policy violations, or recommends corrective action. Some may not even realize those tools qualify as automated decision systems. That is where the risk begins.

The Robot Cannot Make the Call

 Beginning on July 1, 2027, California employers may not rely solely on an automated decision system to discipline or terminate an employee. The law reaches systems that use artificial intelligence, machine learning, statistical modeling, data analytics, and similar technologies to generate scores, classifications, recommendations, or other outputs that influence employment decisions.

So, this is not just about futuristic AI. A system that scores productivity, flags attendance patterns, ranks employees, evaluates call-center performance, identifies suspected misconduct, or recommends corrective action may fall within the law depending on how it works and how the employer uses it.

The practical point is simple: employers will need to know when technology is influencing an employment decision and how much weight leaders are giving it.

A Human Rubber Stamp Will Not Cut It

The most important feature of SB 947 may be its requirement for meaningful human review. If an employer primarily relies on an automated system in deciding to discipline or terminate an employee, a human reviewer must independently corroborate the system’s output using other relevant information.

That could include supervisory evaluations, personnel records, employee work product, witness interviews, peer reviews, or other evidence supporting the decision. The leader reviewing the recommendation cannot simply click “approve” because the software generated a negative score or flagged the employee for action.

The human reviewer needs an independent factual basis for the decision. And if the automated recommendation cannot be corroborated, the employer cannot use it as the basis for discipline or termination. That makes documentation particularly important.

Employees Get to Ask Questions

 The law also gives employees new information rights when an automated system plays a significant role in discipline or termination. Employers will need to provide written notice when the decision is communicated, including information about the role of the automated system and the human review that occurred.

Employees also may be entitled to information about the data used by the system in reaching its result. That requirement should get employers’ attention because many organizations purchase technology without fully understanding what data the system collects, how that information is weighted, or how the software reaches a particular recommendation.

When an employee asks why a system flagged them for discipline, “that is just what the software said” will not be much of an answer.

The Bigger Problem May Be the Software You Already Have

 Employers often think of workplace AI as technology they may decide to purchase in the future. That may be the wrong way to look at the issue.

AI and automated decision-making already may be embedded in payroll systems, applicant tracking platforms, scheduling software, productivity tools, call-center systems, performance-management platforms, security programs, and other applications employers have used for years.

A vendor may have added a scoring, prediction, or recommendation feature without anyone in HR ever thinking of the product as an “AI system.” Employers therefore need to look beyond products marketed specifically as artificial intelligence and identify any technology that produces recommendations or classifications that may influence employment decisions.

In other words, the first step is not deciding whether to buy AI. It is figuring out how much AI you already have.

Start the Inventory Now

 SB 947 does not become operative until July 1, 2027, but employers should not wait until next summer to prepare. The most useful first step is to identify every system that plays a role in hiring, scheduling, attendance, performance management, productivity measurement, discipline, or termination.

Employers should ask what data each system uses, what output it produces, who receives that output, and whether anyone inside the organization can explain how the recommendation was generated. They also should determine whether the underlying facts can be independently verified and whether the leader making the decision could justify it without simply repeating what the software said.

Vendor contracts deserve attention too. If a vendor cannot explain how its system operates or provide enough information for the employer to meet California’s notice and review requirements, that may become the employer’s problem, not the vendor’s.

Leaders and HR professionals also will need training. Human review has to mean something more than approving whatever the computer recommends.

The Bottom Line

 California is not banning AI from the workplace. It is telling employers that when an employee’s job is on the line, a human still has to own the decision.

AI can identify patterns, flag concerns, and make recommendations. But if the best explanation for a disciplinary action is “the system told us to,” the employer may already be relying too heavily on the robo boss.

Want to Get Ahead of This Issue?

 SB 947 is only one piece of California’s rapidly developing workplace AI landscape. Employers also need to think about how AI is being used in recruiting, performance management, workplace investigations, recordings, drafting, policy administration, and everyday HR decision-making.

On Wednesday, November 4 at 10:00 a.m. PT, Joe Beachboard and I will present AI in the Workplace: A California HR Compliance Workshop, a 90-minute practical session sponsored by EmployersLawyer focused on how California employers should actually be using AI now. We will cover what an AI policy should include, where AI is already showing up in HR, the risks employers need to recognize, and the guardrails organizations should put in place before the technology gets ahead of them.

Register here.