Artificial intelligence tools are now embedded in everyday employment decisions — resume screening, candidate ranking, scheduling, performance scoring, and even selection for layoffs. These tools can improve consistency and speed, but recent litigation, expanding state and local rules, and proposed federal legislation have made one point clear: when an AI tool helps make an employment decision, the employer remains legally responsible for the outcome. Managers should understand where the risk lies and take practical steps now.
Why This is a Priority
Courts are testing whether employers — and their AI vendors — can be held liable for biased outcomes.
In a closely watched federal case in California, a court allowed hiring-discrimination claims to proceed on the theory that the software vendor (The AI-utilizing Human Resources system Workday) acted as the agent of its employer-customers when its tool screened, ranked, and rejected applicants — a role that brings the vendor within the definition of “employer” under federal anti-discrimination law — and it certified a collective action. The practical lesson is serious: if the vendor is your agent, the vendor’s violations can become the employer’s violations because the employer’s name is on the rejection notice.
A neutral-looking tool can still create discrimination risk.
These claims rely on a disparate-impact theory, which recognizes that an employment practice is unlawful if it disproportionately harms a protected group and is not job-related and consistent with business necessity, regardless of whether anyone intended to discriminate. That theory remains available under Title VII even though the EEOC has reduced its enforcement under the disparate-impact theory. An AI tool trained on historical data can replicate past bias and produce exactly this kind of disproportionate result.
The regulatory landscape is expanding.
A growing number of state and local jurisdictions now regulate automated employment decision tools. Requirements vary but increasingly include independent bias audits, pre-use notices, risk assessments, and a documented human-review step before the tool drives a significant employment decision. Employers should treat these emerging standards as a baseline, even in locations that have not yet adopted them.
More federal regulation may be coming.
Proposed federal legislation would amend the WARN Act to require layoff notices to disclose when AI was a “substantial factor” in a qualifying mass layoff, including a description of the AI used and an estimate of the resulting job losses. The same bill would create broad whistleblower protections — which could not be waived by contract, policy, or arbitration agreement — for workers who report AI-related legal concerns.
With increased scrutiny in employers’ use of AI tools, employers should take this opportunity to:
-
Identify every AI or automated tool that touches an employment decision — hiring, screening, performance evaluation, scheduling, discipline, and layoff selection. You cannot defend a decision you cannot explain.
-
Understand how each tool works and how it generates its recommendations before relying on it.
-
Have their employment practices reviewed and document a meaningful human-review checkpoint, so a trained person — not the software alone — makes the final call and has the authority to override it.
-
Before adopting, renewing, or expanding any AI tool in the hiring process, consult legal counsel and request the vendor’s bias-audit data and methodology. A vendor’s refusal to share that information is itself a warning sign.
-
Have vendor contracts reviewed to ensure they include meaningful protections and indemnification.
-
Maintain clear, contemporaneous documentation of the criteria used and the reasons for each significant decision.
-
Flag any planned workforce reduction that has an AI dimension, and confirm that anti-retaliation and whistleblower policies are broad enough to cover AI-related concerns.
The Bottom Line
Using an AI tool does not reduce an employer’s obligations under Title VII or other anti-discrimination laws — responsibility for the decision stays with the company. The best protection is to know which tools you use, keep a human decision-maker meaningfully in the loop, vet your vendors, document your reasoning, and seek legal guidance before deploying or renewing tools that influence employment decisions.
