Prepare for Regulations as HR Embraces AI

State and Local Trends

As HR teams embrace new AI technology, they also need to prepare for the inevitable regulations that will follow. Notably, employers are exploring ways to remove human bias by using AI to make data-driven, fair decisions—but such technology can make biased decisions, too, depending on what information is fed into it.

Lawmakers have taken notice. For example, HR professionals in New York City need to be familiar with the city’s groundbreaking new law regulating the use of automated employment decision tools (AEDTs) to hire and promote workers in the city. Employers in the Big Apple must take the following steps if they use such tools to screen workers: 

• Select an independent auditor to perform an annual bias audit of the company’s AEDTs and assess disparate impact based on race, ethnicity and sex.
• Publish the audit results.
• Provide notice to job candidates and employees who reside in New York City about the use of AEDTs.

An initial bias audit must also be performed before using a new AEDT, and the results must be ­published 10 business days prior to using it.

Employers should be prepared to face similar requirements in other locations. For example, California’s AB 331 would have regulated AI use in the state, but the bill was put on hold for now. 

Still, Ebbink notes, California’s bill may serve as a model for future legislation in the Golden State or other locations. Like New York City’s law, the bill was intended to require bias audits or impact assessments for employers that use AI. 

“I anticipate that states are likely to enact legislation to require these impact assessments to ensure there is no disparate impact from the use of AI on protected categories of applicants and employees,” Ebbink says. 

Notably, the California bill would have allowed people to opt out of AI use and request that a real person make decisions.

Federal Focus

HR professionals should also watch for federal legislation. “There is lots of attention and talk about AI at the federal level—and even some legislative proposals,” Ebbink says. For now, he notes, employers may want to review the Biden administration’s Blueprint for an AI Bill of Rights, which identifies equal opportunities in employment as a key area to protect. In addition, the Biden administration has issued an executive order on AI.

“Algorithms used in hiring and credit decisions have been found to reflect and reproduce existing unwanted inequities or embed new harmful bias and discrimination,” according to the blueprint. 

Guarding against any potential adverse impact discrimination seems to be a key focus of federal discussions, Ebbink says. “I think a key question will be whether we see comprehensive federal legislation or a patchwork of state regulation of AI,” he adds.

Don’t forget that existing rules still apply. “The EEOC has been the most active in this space, and employment discrimination laws are probably the first thing that comes to mind,” Ebbink says. 

So, HR professionals should anticipate that federal agencies, including the EEOC, will view artificial intelligence through the same framework as their existing regulations.

“This will continue to be a hot ­topic for federal and state legislation and regulation,” Ebbink says, “so make sure you’re paying attention to what’s happening where you operate.”   

Lisa Nagele-Piazza, SHRM-SCP, is legal content counsel for Fisher Phillips in Atlanta. 

Illustration by Adam Niklewicz for HR Magazine.