While AI use in recruiting and hiring is growing at warped speed, one expert warns about the balance organizations need to strike in pursuing efficiency and creating significant legal risks.
Andrew J. Adams, a partner and chief administrative officer at DarrowEverett LLP, a New York City-based law firm, says AI hiring technology is no longer simply a procurement decision—it is a new legal risk that requires prompt action and strategic forethought.
Few realities underscore that reality more than the lawsuit facing HR tech Workday. This spring, conditional certification was granted in that case (Mobley v. Workday) that opens the door to age discrimination claims on behalf of potentially millions of job applicants who were screened using Workday’s AI recommendation system.
The plaintiff, Derek Mobley, claims that he applied to more than 100 job openings posted by companies using Workday’s AI-powered hiring tools and was rejected each time he submitted an application. Adams says that based on Workday’s discovery, its AI applicant tracking tool screened out one billion applications.
According to the case documents so far, the allegation is that the system’s algorithmic process is biased against applicants older than 40, in violation of the Age Discrimination in Employment Act.
“The court found a common issue among the potential class members: whether Workday’s AI system creates a disparate impact on older applicants,” Adams says.
According to Adams, a common misconception among employers is that if a vendor’s algorithm is screening applicants, liability falls solely on the vendor. The truth is, however, that anti-discrimination liabilities follow the employment decision. As may be the case with Workday, the use of a hiring vendor tool that creates discriminatory outcomes may result in joint liability for the employer along with the vendor. He says vendor “certifications of fairness or compliance alone” cannot serve as a legal shield.
“Many employers assume that if the AI came from a third-party vendor, the legal risk stays with the vendor,” Adams says. “That’s not how employment law works. Employers remain responsible for the hiring decisions being made in their organizations, even when technology is involved.”
Adams explains that what makes the case important are a trio of critical issues that HR and hiring pros must consider when using AI:
- For one, Adams explains that the court accepted the argument that Workday might be liable as an “agent” of the employers that use its tools, establishing the crucial distinction between simple tools (such as email or spreadsheets) and AI-powered systems that perform decision-making tasks. “Any vendor whose AI tool is used to make employment decisions may have joint liability under Title VII, the ADEA and the ADA,” Adams says.
- No allegations of intentional discrimination are required to move forward with discrimination claims. An AI tool that produces “statistically skewed outcomes” can make an organization liable, irrespective of any intention, Adams says.
- Finally, Adams says, the court ordered Workday to provide a list of customers who turned on the AI feature, indicating that there could be claims of discrimination against employers that receive scoring or ranking from the system.
As for ultimately avoiding being on the wrong side of this decision, Adams says HR leaders should consider AI hiring as part of a broader approach to enterprise risk management. Along those lines, he offers several immediate suggestions:
- Run “privileged bias audits” of all existing AI-powered screening and recommendation tools used by the company. Also, Adams says, audit any algorithms under the supervision of legal counsel in a manner that preserves attorney-client privilege.
- Keep meaningful human oversight at critical junctures. According to Adams, the AI system should inform about the hiring of applicants and should not serve as the ultimate decision-maker. “When an algorithmic tool rejects a qualified applicant because of an employment gap or an unconventional resume, there should be someone to give a human perspective,” he says.
- Audit vendor contracts. Here, Adams says, employers must insist that vendors provide documentation on the bias testing, ensuring there are contractual guarantees related to nondiscrimination and data transparency, and require cooperation with independent audits.
- Constantly analyze hiring results for any trends, including those affecting protected classes, including age, race, gender and disability status. He says HR must help employers consider any significant disparity as a red flag that needs an immediate investigation and potential remediation.
- Stay ahead of the regulatory patchwork. Adams notes that several states, including California, Illinois and Colorado, have already implemented laws covering the use of AI in hiring. “More states will follow,” he says.
“The trend is clear,” says Adams, noting that courts are willing to hold both vendors and employers responsible for algorithmic outcomes.
“Organizations that act now to audit their tools, strengthen vendor oversight, maintain human involvement and document good faith compliance efforts will be far better positioned to manage the legal, financial and reputational consequences of this new, emerging enforcement era,” Adams says.
Credit: Source link
