AI Hiring Tools and the Employer’s Proof Problem
Most employers using AI in hiring assume the vendor carries the legal risk. The vendor agreement probably says otherwise, and California’s regulatory framework makes it even clearer: the employer owns the workflow.
The California Civil Rights Department (CRD) has issued regulations addressing Automated Decision Systems (ADS) used in employment decisions. These regulations signal that when AI tools are used to screen, rank, or evaluate applicants, the employer, not the software provider, is accountable for what the system produces. The CRD’s framework follows FEHA’s existing structure: if the outcome disparately affects a protected group, the employer must justify the tool’s use on job-related, legitimate business grounds.
California’s privacy framework adds a second layer. Under the California Consumer Privacy Act (CCPA) as amended by the California Privacy Rights Act (CPRA), employers may be required to conduct a risk assessment before deploying Automated Decision-Making Technology (ADMT) that makes, or substantially contributes to, a significant decision affecting employees or applicants. Screening tools that rank or eliminate candidates likely qualify. A risk assessment that was never done is a document gap that surfaces in litigation.
Vendor contracts often include indemnification provisions, but those provisions rarely transfer full liability. They do not guarantee that the employer’s disclosures, authorizations, and documentation practices were adequate. They do not prevent a plaintiff from naming the employer in a disparate-impact claim. And they are often silent on the question of whether the employer audited the tool before deploying it.
The proof problem is practical: can the employer articulate, in human-readable terms, what criteria the tool used, how those criteria were validated as job-related, and how the tool was monitored over time? If the answer is no, the system itself becomes a legal issue before anyone argues about whether a specific hiring decision was fair.
When vendor reliance, risk assessments, and bias audits do not align, AI hiring tools become a California employer’s liability, not just a procurement decision.
If you want to know where your business stands, contact Michael Trust Law for a no-charge initial consultation. The facts determine what needs to be addressed — and how much of a conversation that takes.
Disclaimer:
This post shares general information based on common patterns I see in California workplaces. It is not legal advice, does not create an attorney-client relationship, and outcomes depend on specific facts — no lawyer can guarantee a result. Past results do not guarantee or predict future outcomes. AI may have been used to create this post. All content reviewed by a CA attorney before publication.