Contingent labor programs have long relied on a workable fiction: the staffing supplier is the employer of record, the client company is merely the beneficiary of services, and the two remain legally distinct. That fiction is now under pressure. As suppliers embed AI sourcing tools, resume screeners, and automated matching engines into their fulfillment workflows, the question of who actually controls the hiring decision becomes harder to answer. Under the common-law test articulated by the Supreme Court in Nationwide Mutual Insurance Co. v. Darden, 503 U.S. 318 (1992), that control question is the pivot on which co-employment liability turns.
The Darden test enumerates roughly a dozen factors drawn from agency law, including the skill required, the source of instrumentalities, the duration of the relationship, the method of payment, and the tax treatment of the worker. Courts and agencies, including the EEOC in its Enforcement Guidance on Contingent Workers (Notice 915.002, Dec. 3, 1997), have consistently treated the right to control the manner and means of work as the most heavily weighted factor. When a client company shapes how a supplier finds, filters, and forwards candidates, that client begins accumulating indicia of control long before the worker sets foot on site.
How AI Sourcing Tools Amplify the Control Factor
AI-driven candidate matching tools do not operate in a vacuum. They ingest client-supplied requisition data, historical hire patterns, performance signals, and often explicit rejection feedback from the client's hiring managers. When a client instructs a supplier to configure its algorithm around a specific competency profile, tune its scoring against past client hires, or exclude candidates flagged by a client-mandated assessment vendor, the client is exercising direction over the means of selection. That is precisely the type of instruction the Darden analysis captures. The more the supplier's AI is trained, calibrated, or gated by client input, the harder it becomes to argue the supplier alone controls the hiring process.
Regulatory scrutiny compounds this risk. NYC Local Law 144 requires bias audits and candidate notice for automated employment decision tools used to screen candidates for positions in the city, and the statute reaches employers and employment agencies alike. Illinois' Artificial Intelligence Video Interview Act and the state's broader AI amendments to the Illinois Human Rights Act (HB 3773, effective January 1, 2026) impose obligations on employers using AI in recruitment. Colorado's SB 24-205, the Colorado AI Act, will impose developer and deployer duties around high-risk AI systems in employment starting in 2026 [VERIFY effective date given recent amendments]. Under the EU AI Act, recruitment and candidate evaluation systems are classified as high-risk, triggering conformity assessment and human oversight obligations for deployers. If a client is deemed a joint employer or a co-deployer of the supplier's AI, these obligations attach directly to the client.
Practical Steps to Contain the Exposure
Clients that want the efficiency of AI-enabled staffing suppliers without inheriting employer status should treat supplier tooling as a governed part of their own AI inventory. Contracts should require suppliers to disclose which automated tools are used at each stage of sourcing and screening, provide bias audit results consistent with the Uniform Guidelines on Employee Selection Procedures and applicable state law, and warrant compliance with jurisdiction-specific notice and consent requirements. Clients should avoid dictating algorithmic configuration, avoid supplying rejection rationales that retrain supplier models on client-specific criteria, and avoid embedding client personnel in the supplier's screening decisions. Where client input is unavoidable, document the business necessity and preserve the supplier's independent judgment on final candidate presentation.
The safest posture is to assume that a plaintiff's lawyer, an EEOC investigator, or a state attorney general will one day map every touchpoint between client and supplier against the Darden factors. AI creates more touchpoints, not fewer, because algorithmic systems generate logs, feedback loops, and configuration artifacts that make control visible in ways handshake staffing arrangements never did. Suppliers and clients who validate their AI governance jointly, document the boundary between them, and treat co-employment risk as an AI compliance question rather than a purely contractual one will be far better positioned when that mapping exercise begins. Don't trust the supplier's assurances. Validate them.
