California regulations going into effect on October 1, 2025, make clear that employers’ use of artificial intelligence in personnel practices can lead to liability under state fair employment laws.

Here are two key takeaways:

• The regulations clarify that an employer’s use of an automated-decision system or selection criteria that discriminate against applicants or employees on a basis protected by California law can give rise to employer liability. In short, AI engaging in unlawful discrimination may be attributed to the employer using the AI.
• Employers may run systems using AI through anti-bias testing or take similar proactive efforts to avoid unlawful discrimination claims or more effectively defend them.

The information presented is not intended to be, and does not constitute, “legal advice.” Because each situation varies, and only brief summary information is provided here, you should not use this information as a basis for action unless you have independently verified with your own counsel that it applies to your particular situation.

The newly introduced “No Robo Bosses Act” seeks to regulate the use of AI in the workplace and prevent automated decision-making processes in employment decisions. For employers, understanding the implications of this bill is essential to maintaining compliance and minimizing legal risks, especially as AI becomes increasingly integrated into workplace practices. If enacted, SB 7 would have an immediate impact on employers using AI for workforce management.

As businesses increasingly use AI to boost efficiency and productivity in the workplace, there are currently no safeguards to prevent machines from unjustly or illegally impacting workers’ livelihoods and working conditions. However, the California Civil Rights Division is currently considering similar regulations regarding automated decision-making systems (“ADS”).

The bill seeks to stop California employers from relying solely on ADS for critical workplace decisions—like hirings, terminations, disciplinary actions, and promotions. Under SB 7, “worker” means any person who is a job applicant to, an employee of, or an independent contractor providing service to, or through, a business or a state or local governmental entity in any workplace. Therefore, it would apply to job applicants and independent contractors, not just employees.

The bill defines an ADS as any computational process that uses machine learning, statistical modeling, data analytics, or artificial intelligence to produce simplified outputs—such as a score, classification, or recommendation—that assist or replace human decision-making and materially impact individuals. However, common workplace technologies were omitted from the definition, such as spam email filters, firewalls, antivirus software, and identity management tools.

In terms of enforcement, SB 7 would prohibit an employer from discharging, threatening to discharge, demoting, suspending, or in any manner discriminating or retaliating against any worker for taking certain actions asserting their rights under this Act. In addition, it would require the Labor Commissioner to enforce its provisions and would authorize a public prosecutor or any worker who has suffered a violation or their representative to bring a civil action.

If the bill is signed into legislation, California employers would need to take steps to ensure compliance in several ways, including but not limited to:

1. Human Oversight Requirement: Employers would need to ensure that any employment decision influenced by AI also passes through meaningful human review. AI cannot be the sole source of employment-related decisions under SB 7.
2. Transparency: Employers would be required to notify employees, in writing, when AI tools are being used in decision-making processes. The bill would further require the employer to maintain a list of all ADS currently in use and would require the notice to include the updated list. This could necessitate revising employee handbooks and updating onboarding procedures to include AI usage disclosures.
3. Data Privacy: Employers must ensure that data used in AI-decision making is handled responsibly and securely.
4. Predictive Behavior Analysis: If AI-driven decisions result in disparate impacts on protected classes, employers could face discrimination claims. Implementing robust auditing processes of ADS and conducting regular bias testing will be essential to mitigate this risk.
5. An ADS cannot obtain or infer information about an employee’s immigration status, ancestral history, health history, credit history, or other statuses protected by state law: Employers would need to implement protections in their ADS in this regard.
6. Enhanced Training for Employees: Employers may consider additional specific training on AI and provide procedures to oversee AI-driven decisions.
7. Appeals: Employees would have an opportunity for appealing decisions made by an ADS. Employers would need to work on an appeals process.

As AI continues to gain traction across industries to improve efficiencies and reduce costs, it is important for California employers to proactively assess their current practices and implement safeguards if they use AI in employment decisions.

The information presented is not intended to be, and does not constitute, “legal advice.” Because each situation varies, and only brief summary information is provided here, you should not use this information as a basis for action unless you have independently verified with your own counsel that it applies to your particular situation.

AI has already changed the way many access information, perform work, and do business. However, it is important to remember “it’s a tool,” “garbage in – garbage out,” and “check your work.”

Courts throughout the nation have been subjected to filings generated by AI. AI has a proclivity to “hallucinate” legal propositions and case law citations. As a result, Courts in multiple jurisdictions have issued Standing Orders on the use of AI and a few have banned its use entirely. Courts, constrained by research attorney’s hour limitations, readily rely on the adversarial process to point out problems in a party’s brief or reasoning. However, more and more courts have been faced with incurring significant time trying to run down non-existent cases, case citations and quotations, and statutes conjured by AI. Thus, many Courts now impose Declarations under penalty of perjury by attorneys and/or parties about their use of
AI in their filing, that they have checked all citations for validation and that they have not misrepresented the law. This allows the Court to sanction attorneys and parties if the cases, citations and/or statutes cited are non-existent or misrepresented.

AI is only a tool. There was an immigrant engineer who learned the old-school way to make engineering calculations on intricate designs. He went on to engineer many earthquake resistant hospitals throughout the world. He became a senior engineer at a large firm. Then the computer era began. His junior associates, who had only learned on computers, would bring engineering designs to him for his final review. Inevitably one or more of the elements of the design were wrong. He would ask the junior associate, why did you think this was correct?” Inevitably, the junior associate would respond, “because the computer said it would work.” The computer, just like AI, is only a tool. Garbage in – garbage out.

When making legal decisions about your life and business do not rely on AI – at least not yet. Have a qualified attorney “check your work.”

The information presented is not intended to be, and does not constitute, “legal advice.” Because each situation varies, and only brief summary information is provided here, you should not use this information as a basis for action unless you have independently verified with your own counsel that it applies to your particular situation.