Labor and employment attorneys warn of AI risks in hiring and personnel decisions

Listen to this article

A late-2024 survey by Insight Global found that 99% of hiring managers report using artificial intelligence (AI) in some capacity during the hiring process. As employers increasingly use AI tools in hiring, performance management and other personnel decisions, labor and employment attorneys say organizations also need to understand the legal, privacy, and compliance risks that can accompany them.

“I think the biggest risk right now is over-reliance or overly trusting AI to make employment decisions,” said Ibrahim Tariq, a partner and member of the Labor and Employment Practice Group and Artificial Intelligence Industry Team at Harris Beach Murtha. “The reason for that being at the end of the day, any decision around hiring, advancement, promotions, compensation, that’s a human choice.”

From what Tariq has seen, where AI tends to be a problem for HR and for management is when a company is relying solely on the output generated by AI. For example, a company might upload employee sales numbers, performance metrics, or other workplace data into an AI platform and then use the output to help determine promotions, raises, or disciplinary actions.

“The challenge becomes whether the AI has properly accounted for bias and discrimination risks and whether it is inadvertently treating employees of a certain characteristic worse than others,” he said.

Tariq pointed to a widely publicized Equal Employment Opportunity Commission (EEOC) lawsuit involving a tutoring company whose AI hiring tool allegedly screened out older applicants.

While he described that case as an extreme example, he said similar concerns about bias and unintended discrimination can arise on a smaller scale whenever employers rely too heavily on AI-driven decisions.

Tariq said employers should be especially cautious when using AI in hiring, firing and compensation decisions, which he described as among the highest-risk areas for employment-related claims.

“Blind reliance on whatever the AI said goes, that’s not a defense to an employment law claim,” he said, adding that companies also need to think carefully about what information employees are entering into AI systems, particularly when it involves confidential or sensitive company data.

As AI adoption accelerates, Tariq said organizations should establish clear internal policies governing acceptable AI use and provide at least basic training for managers and HR professionals.

He also noted that AI regulation continues to evolve at the state and local levels, creating a patchwork of rules that employers will need to monitor closely, particularly if they operate across multiple jurisdictions.

Kim Harding, an attorney in the Labor & Employment group and the leader of the Labor Relations team at Nixon Peabody, said employers should think about AI not simply as a tool, but as a third party that may have access to sensitive company and employee information.

“We really need clients to understand that AI is effectively a third party,” Harding said. “It becomes easy to treat it like just a machine or just a tool.”

Harding said one of the biggest legal concerns involves discrimination exposure and the potential for disparate impact claims tied to AI-assisted employment decisions. But she emphasized that privacy, confidentiality, and vendor oversight are equally important considerations for employers adopting AI systems.

Harding said organizations should carefully evaluate what information is being entered into AI platforms and whether confidential, proprietary, or employee-related data could be exposed through those systems. She also encouraged employers to scrutinize the contracts and terms associated with AI vendors rather than simply accepting standard agreements.

“We see too often people just sign the standard model contract or the boilerplate contract,” Harding said. “This, like any other area, is an opportunity to really understand what you’re signing.”

As a starting point, Harding recommended that employers first determine what AI tools employees may already be using, whether officially approved or not, before implementing broader governance policies. From there, companies should establish clear guidelines around authorized AI use, provide training and ensure HR professionals understand both the capabilities and limitations of the technology.

Harding also stressed the importance of human oversight and cautioned employers against simply accepting AI-generated outputs at face value. Employers, she said, need to understand how decisions are being made and be prepared to defend them if challenged.

Finally, Harding noted that successful AI governance efforts often involve collaboration across departments, including HR, legal, IT and executive leadership.

“Many voices at the table,” she said, are critical as organizations navigate what remains a rapidly evolving area.

Benjamin E. Mudrick, a partner and leader of the Labor & Employment practice at Harter Secrest & Emery, said one of the biggest challenges with AI in employment decisions is that workplace issues often require nuance and individualized judgment.

“Employers have to treat their employees in a consistent and equitable manner, but at the same time they have to treat every situation on an individual basis,” Mudrick said.

Mudrick said AI systems are designed to generate consistent outputs based on large sets of historical data, but that can create problems when employers rely too heavily on automated recommendations in areas such as hiring, promotions or discipline.

He said AI may overlook circumstances that are not reflected in the data being analyzed or unintentionally reinforce discriminatory patterns already embedded in prior decision-making.

Mudrick also noted growing concerns around AI-assisted hiring tools, pointing to jurisdictions such as New York City that now require bias audits for certain AI-driven employment systems. Employers, he said, still face the same legal scrutiny for discrimination claims even when decisions are made or supported by AI.

“The decisions that are made through AI still face the same legal scrutiny that any other employment decision makes,” he said.

Mudrick also cautioned employers to think carefully about confidentiality and discoverability when using AI platforms. Entering sensitive employee or health-related information into AI systems, for example, could create privacy risks under laws such as the ADA or HIPAA, while questions posed to AI chatbots may later become discoverable in litigation.

At the same time, Mudrick said AI use in the workplace is only continuing to expand.

“We’re seeing more companies use it and more products being offered,” Mudrick said. “And I think that’s a good thing, but employers need to go in with their eyes wide open.”

Caurie Putnam is a Rochester-area freelance writer.