A bill approved by the California Legislature would place a concrete limit on the use of artificial intelligence in the workplace: companies would not be allowed to rely exclusively on automated systems to decide whether an employee should be fired or disciplined if the proposal becomes law. The change could reshape how businesses document and oversee employment decisions made with the help of AI.

California wants a human between AI and a firing decision

California approved SB 947, known as the No Robo Bosses Act of 2026, on August 31, a bill designed to prevent automated decision systems from having exclusive control over decisions to fire or discipline workers.

The measure passed the Senate by a 28-10 vote, after clearing the Assembly 53-14. It is now on Governor Gavin Newsom’s desk and still needs his signature to become law.

The central idea is straightforward: a company could continue using artificial intelligence to support internal processes, but an automated system could not be the sole authority determining whether an employee is fired or disciplined.

The difference between assisting and deciding

That distinction matters because it changes the debate over workplace automation. A tool can analyze data, identify patterns or flag potential problems without necessarily being given the authority to make the final decision.

SB 947 establishes precisely that boundary. When an automated system is involved in a firing or disciplinary decision, there must be human oversight and verification.

This creates an important separation between using AI as a decision-support tool and allowing an algorithm to assume direct responsibility for a decision that could alter someone’s career and livelihood.

What the bill would require companies to do

Corporate environment with a human manager reviewing a decision generated by artificial intelligence

The bill seeks to keep final responsibility for employment and disciplinary decisions in human hands.

The proposal does not treat artificial intelligence as something that must be removed from businesses. Instead, it focuses on limiting the authority of automated systems when their decisions can directly affect a person’s professional life.

In practice, companies using these systems to support firing or disciplinary decisions would have to incorporate human review and verification into the process. The goal is to prevent a recommendation produced by an algorithm from automatically becoming an employment consequence.

Transparency is another important part of the proposal. Workers would have to be notified when an automated system was used in a firing or disciplinary decision. The bill also includes mechanisms related to the information used by the system to reach its recommendation.

The problem is not only technological

An automated decision can appear objective because it is presented through data, scores or classifications. But that does not necessarily mean the result is correct or appropriate for an individual employee’s circumstances.

This is where human review becomes important. The bill is based on the idea that a decision affecting someone’s salary, employment and career should not be reduced to an automated software output.

The proposal also reflects a broader debate over how much autonomy companies are giving AI systems. As more tasks are performed without direct human intervention, the need to establish clear boundaries around those decisions becomes increasingly important.

Why the decision matters to workers

The debate stops being abstract when artificial intelligence becomes involved in decisions that can determine whether someone continues receiving a paycheck.

The debate surrounding SB 947 has gained momentum amid reports of automated workplace systems producing problematic decisions, including cases in which workers were allegedly terminated incorrectly. The office of Senator Jerry McNerney has also cited the growing use of tools known as bossware, which are used to monitor and manage workers.

That is the most consequential aspect of the proposal: California is treating workplace AI not merely as a productivity tool, but as a technology capable of directly affecting the relationship between employers and employees.

When an algorithm affects a career

For workers, the difference between a recommendation and a decision is enormous.

A system can flag a particular employee’s behavior or performance. But turning that assessment into a termination involves context, employment history, individual circumstances, and legal and managerial responsibility.

California’s proposal is designed to prevent that final step from disappearing behind an automated decision.

The debate is unfolding just as major companies are expanding their use of AI agents to perform tasks previously handled by employees. Notícia Tech has already examined this shift in its report on how Meta plans to replace employees with AI agents, showing that the debate over workforce automation extends far beyond human resources departments.

For companies, AI in HR may require greater accountability

Executives analyzing artificial intelligence data during a human resources meeting

The legislation could increase the need for documentation and governance around AI systems used in employment decisions.

If SB 947 is signed into law, companies using automated systems in employment processes will need to pay closer attention to how those tools participate in decisions.

That could mean more structured internal processes, records documenting decisions, clear identification of when AI was used, and defined responsibility for who performs the human review.

The change also reinforces a discussion already emerging in other areas of enterprise technology: the more autonomy a company gives to AI systems, the greater the need for governance, oversight and control mechanisms.

That concern is already visible across the corporate market. As systems become capable of performing tasks with greater autonomy, businesses are considering new forms of control. Notícia Tech has previously reported how Microsoft warns that AI agents require new governance in businesses, a discussion directly related to the type of oversight SB 947 seeks to establish in the workplace.

The cost of automated decision-making

For a business, automating a decision can save time and reduce costs. But an automated decision that produces the wrong outcome can create another category of costs: disputes, rework, loss of trust and regulatory risk.

The challenge becomes finding the right balance between productivity and accountability.

The question is no longer simply whether a company can use AI. It is which decisions it can entrust to AI, how much autonomy the technology should have, and who is accountable for the outcome.

California could be setting a new boundary for AI in the workplace

Employee leaving a corporate office while managers review an AI-assisted employment decision

The debate over workplace AI is moving beyond productivity and toward accountability for decisions that directly affect people.

SB 947 does not yet mean California has definitively prohibited companies from using AI in employment decisions. The bill still needs the governor’s signature before its provisions can take effect.

But its approval by the Legislature already signals an important shift in the regulatory debate: the question is no longer only what AI can do, but what companies should allow it to do on its own.

If the bill becomes law, California could become the first U.S. state to establish this type of specific requirement for firing and disciplinary decisions involving automated systems.

For companies expanding their use of AI, the message is significant. Automation can perform tasks, analyze information and recommend actions, but decisions with a direct impact on people may require an additional layer of accountability.

This shift also helps explain why AI governance is becoming a central part of enterprise AI adoption. As systems move beyond answering questions and begin influencing real-world decisions, human oversight, documentation and accountability are no longer merely compliance details. They are becoming part of the architecture of automation itself.

The next step, therefore, is not only to watch whether California Governor Gavin Newsom signs SB 947 into law. It will also be important to see whether other jurisdictions and companies begin establishing similar boundaries around how far artificial intelligence should be allowed to go when making decisions about people’s jobs.