California Bans Companies From Using AI to Fire Workers Under Senate Bill 1024

On September 28, 2026, California Governor Gavin Newsom reportedly signed Senate Bill 1024, a measure that would bar companies from using artificial Intelligence as the sole basis to fire workers. The supplied details are not yet independently confirmed, so this report treats the legislation and penalties as reported facts rather than settled legal guidance.

Before Senate Bill 1024: AI Could Influence Automated Employment Decisions

Before the new measure, companies could use automated employment decision-making tools to assess productivity, attendance, performance indicators, or workplace conduct. Those systems could influence managers even when the final termination decision was formally made by a human. The practical concern was that an algorithm could become the decisive factor without a worker receiving a clear explanation or meaningful review.

The legislation reportedly amends the California Labor Code and targets automated employment decision-making tools. Its stated purpose is not to prevent companies from using AI in employment altogether. Instead, it draws a line between using software as an input and allowing software to become the sole reason for termination. That distinction matters for businesses building AI-assisted human resources systems. A manager who reviews an automated recommendation, checks the underlying records, and makes an independent decision would face a different compliance question from a company that automatically dismisses workers after a software-generated score falls below a threshold.

California’s Catalyst: Senate Bill 1024 Changes the Rule

Senate Bill 1024 reportedly sets January 1, 2027, as the point by which the new framework for automated employment decision-making tools must be addressed. The measure also reportedly introduces auditing requirements affecting major California-headquartered technology employers, including Alphabet Inc. and Meta Platforms Inc. Those requirements would place internal data practices, model outputs, and termination workflows under closer scrutiny.

The California Chamber of Commerce opposed the bill, citing compliance burdens for approximately 1.5 million registered businesses in the state. That objection reflects the uneven cost of regulation: a global platform can assign legal and engineering teams to an audit, while a small employer may rely on a third-party payroll or workforce-management service. The law reportedly also establishes financial exposure. Violations of the automated termination ban carry civil penalties starting at $5,000 per affected employee, with enforcement assigned to the Department of Industrial Relations. For a business with a large workforce, the per-worker structure could make a single flawed automated process substantially more expensive than a conventional compliance review. For more detail, see OpenAI Blog.

After the Ban: Companies Must Keep Human Judgment in the Loop

The immediate change is operational rather than cosmetic. Companies will need to document how AI tools contribute to employment decisions, identify who makes the final call, and preserve evidence that a human reviewed the case. That may push employers toward approval workflows, audit logs, model monitoring, and clearer explanations for workers.

The measure does not make every use of AI in human resources illegal. It focuses on the specific situation in which artificial Intelligence becomes the sole basis for firing someone. Employers can still face difficult questions when a human signs off on an AI recommendation without genuinely evaluating it. The same principle is appearing across workplace technology: software can recommend, rank, or flag, but accountability remains with the organisation. Even a simple tool used for Using Remote Desktop can create records that employers must handle carefully when automated monitoring influences discipline.

Before And After: Automated Termination Rules

The law’s success will depend on how regulators define genuine human judgment. A signature alone may not be enough if an employer simply approves every automated recommendation, while overly rigid rules could prevent companies from using legitimate productivity and safety tools. For more detail, see VentureBeat AI.

AI Employment Compliance Before January 1, 2027

The next stage will centre on implementation, audits, and enforcement guidance. Businesses will need to map every automated employment tool, identify whether it can trigger termination, and establish a review process that produces a defensible record.

If you run a large technology company, pick a formal audit programme before January 1, 2027. If you operate a smaller business, pick documented human approval and vendor checks before relying on automated employment scores. If you are a worker, pick written explanations and review rights whenever an AI-linked decision affects your job.

FAQs

Can companies still use AI to monitor workers?

Yes, the reported ban targets AI as the sole basis for firing workers, not every form of workplace monitoring or workforce analysis.

What is Senate Bill 1024?

Senate Bill 1024 is the reported measure that amends the Labor Code to regulate automated employment decision-making tools from January 1, 2027.

How much can a violation cost?

Civil penalties reportedly start at $5,000 for each affected employee.

Who enforces the rule?

The Department of Industrial Relations is responsible for enforcing the reported automated termination penalties. The next test is simple: AI may recommend a decision, but companies must prove that a person actually made it.