Employers using AI to screen resumes, assess interviews or rank candidates are being warned that the tools create direct litigation risk, not just procurement or IT risk.
Disparate-impact claims can arise when training data, model design or proxy variables such as ZIP code and school systematically disadvantage protected groups, even without intentional discrimination.
ADA obligations still rest with employers when AI assessments screen out applicants with disabilities, requiring alternative evaluation methods rather than reliance on vendor claims of compliance.
Black-box vendor tools do not shift liability: employers need validation studies, adverse-impact testing, audit records and human review to explain challenged hiring decisions.
Federal, state and local rules are evolving separately, pushing employers to inventory AI tools, monitor multijurisdictional requirements and build compliance into hiring workflows from the start.