California is doing what it can to protect you from judgmental AI tools that may be causing more harm than good. The California Civil Rights Council is all set to bring its California AI anti-discrimination regulations to the forefront of the conversation surrounding the use of AI and automation in recruiting and employment.
The California AI hiring rules are not set in stone yet, but progress has been made in establishing a more robust way to regulate how employers use these systems independently without supervision. These automated systems have been accused of being discriminatory and ineffective, and employers must set their own guidelines in place while relying on such tools on the matter of hiring. Let’s understand what the California AI discrimination rules have to say.
California’s AI discrimination rules will set the guidelines for employers who want to embrace AI in hiring. (Image: Pexels)
California AI Anti-discrimination Regulations: What Changes Do They Propose?
The Civil Rights Council, part of the California Civil Rights Department (CRD), has shown its support for the regulations based on California’s new “Employment Regulations Regarding Automated-Decision Systems.” The new rules are still pending final approval—no deadlines have been set for when it might go into action—however, they are expected to go into motion in 2025.
Automated-Decision Systems (ADS) are a new advancement of AI technology and other adjacent mechanisms that make or support decision-making in the realm of employment. These systems simplify the process of hiring by taking up a large chunk of the manual labor, but their independent decision-making capabilities cannot be trusted without supervision.
The terms used in the bill are similar to Colorado’s AI Discrimination Law and its stance on algorithmic discrimination, and the emphasis on ensuring that AI tools—used increasingly for hiring, promotions, and performance evaluations—do not perpetuate bias or violate the discrimination laws of the region.
When Will the California Regulators Adopt the AI Rules?
The California Civil Rights Council (CRC) has already approved the regulations, but they still need to be signed off by the Office of Administrative Law (OAL) before they can go into effect. They should be fully approved in the coming months, so employers in the region can expect California’s AI anti-discrimination regulations to govern them by the end of the year.
It could go into effect as early as June, as the California AI hiring law is nearing approval. So, if you run a business in the region, you should start making changes in preparation for the new rules.
What Employers Need to Know About Automated Decision Systems and CA Regulation in 2025
The California AI anti-discrimination regulations don’t ban ADS but instead outline the rules for their uses and the considerations to be kept in mind by employers. Recruiters cannot allow any room for bias, even if it’s the system that’s being discriminatory, and even third-party ADS suppliers can be held accountable for any hits or bias or unfair decision-making.
Here’s everything you need to know about the California AI hiring rules and their guidelines for using AI systems in employment decisions.
The California Regulations Ban AI-Driven Discrimination
- Employers cannot use ADS to discriminate based on protected traits like race, sex, age, disability, religion, or national origin
- This covers both disparate treatment (intentional bias) and disparate impact (unintended harm)
- Technology cannot be used as an excuse for unlawful practices, so the employers are to be held accountable for what the technological tools do
- For example, an AI rejecting more female candidates could violate the rules, even unintentionally
Third Parties to Be Held Accountable for ADS
- Not only can the employer be held liable for discrimination, but also potentially third parties who design or implement these systems could be implicated
- An organization must be able to prove that rigorous anti-bias testing was conducted to assess the software
- The final regulations do not include previously proposed language that would have explicitly created third-party liability for the mere design, development, advertising, promotion, or sale of discriminatory AI systems. Now, they will be held accountable for “aiding and abetting” rather than for creating the system
Record-Keeping for Accountability
- Employers must keep AI-related records for four years, up from two in earlier drafts
- This includes job applications, personnel files, and ADS data
- By ensuring data is stored, employers can ensure transparency and enforcement to show what procedures were used in their decision-making
- It also ensures that evidence is available for regulators and employees in disputes
Balancing Business Needs with Fairness
- Use of ADS systems can be inherently discriminatory, even if that isn’t the intention of use
- For example, measuring skills such as dexterity and reaction time when they are not needed for a job might be discriminatory to those with physical limitations, as they might be perfectly able to showcase skills that are actually needed
- ADS use is only permissible if it is job-related or tied to “business necessity” and should include a mechanism for accommodations or adjustments
- Employers must show that no less discriminatory alternative exists to defend their use of ADS tools
- Employers that can provide evidence of fairness, like anti-bias efforts, will be better able to defend their use of such software. Without such efforts, claims of necessity may falter in legal challenges
Protecting Privacy in Medical and Psychological Assessments
- Existing rules that prohibit employers from asking job applicants about their medical or psychological history now include inquiries made using ADS
- ADS cannot conduct medical or psychological evaluations or “unlawful inquiries” through other tests to acquire this information
- This includes personality tests or gamified screens assessing mental/physical traits
- The rule protects employees from invasive or unnecessary probes
- It adapts FEHA’s existing limits to modern AI tools
The California AI Anti-discrimination Regulations Extend to Criminal History
- California law already mandates that before employers deny a job applicant based on their criminal record, they must conduct an “individualized assessment”
- This assessment will be used to determine if their specific conviction history has a direct relationship with the job they are being hired for
- As a result, ADS alone cannot replace individualized assessments of criminal history with a blanket exclusion of all candidates with a record
- The regulations uphold California’s requirement for human judgment and ensure fairness in evaluating past records
As California Regulators Adopt AI Rules, More States Will Follow
The California AI anti-discrimination regulations have attempted to create a reliable system around the use of AI and ADS tools in making hiring, promotions, and other employment decisions in a workplace. These rules protect workers and their right to be hired on the basis of their qualifications and eliminate potential room for discrimination based on some protected factors.
Once we see more employers start relying on these automated systems, it is likely that these rules and guidelines will need to be updated to keep up with the changing times. As of now, employers that choose to use ADS need to evaluate the system before employing them and ensure there is some degree of human supervision throughout the process.
It is best to test for any signs of discriminatory qualities in the tools and rectify them at the earliest to ensure they are in the green on its use.
Subscribe to The HR Digest so we can help prepare for incoming laws and regulations in advance, so you never have to face a challenge in overseeing your workforce correctly.