Civil rights agency drops a key tool used to investigate workplace discrimination

Key Points

• The U.S. Equal Employment Opportunity Commission (EEOC) has decided to stop investigating complaints based on "disparate impact liability," which targets policies that may unintentionally discriminate against certain groups.
• This decision aligns with President Trump's executive order aimed at reducing the use of disparate impact theory in civil rights enforcement, citing it as a barrier to meritocracy.
• The shift has raised concerns among critics who argue it undermines a crucial tool for identifying discrimination, particularly with the increasing reliance on artificial intelligence in hiring processes.
• Due to the new policy, workers whose complaints are based solely on disparate impact will have to pursue their cases in court, eliminating the benefit of a publicly-funded investigation by the EEOC.

The U.S. civil rights agency responsible for enforcing worker rights will stop investigating complaints about company policies that don’t explicitly discriminate but may disproportionately harm certain groups, according to an internal memo obtained by The Associated Press.

The memo, emailed to all area, local and district office directors of the U.S. Equal Employment Opportunity Commission on Sept. 15, says that the agency will discharge by Tuesday any complaints based on “disparate impact liability," a legal concept that argues that even if a policy looks fair on the surface, it can still be discriminatory if it creates unnecessary barriers that make it harder for certain groups of people to succeed.

The EEOC's decision to drop such cases aligns with President Donald Trump’s April executive order directing federal agencies to deprioritize the use of disparate impact in civil rights enforcement because it encourages the assumption that any racial imbalance in the workforce is a result of discrimination, which creates undue burden on businesses.

The move marks a significant shift in EEOC enforcement, and critics say it weakens an effective legal tool used to root out workplace discrimination. That's especially true when it comes to addressing algorithmic bias as more employers rely on AI in the hiring process.

Since AI draws on large amounts of data to generate results, it can replicate the patterns of inequality even if it's not programmed to do so. In one infamous example, Amazon developed a resume-scanning tool to recruit top talent, but abandoned it after finding it favored men for technical roles — in part because it was comparing job candidates against the company’s own male-dominated tech workforce.

“As AI is becoming more and more popular, it’s particularly important that we have the disparate impact tools available to be able to police it and make sure it’s not being used to resegregate the workforce,” said civil rights and plaintiff-side employment attorney Christine Webber.

Historically, discrimination claims invoking disparate impact have been less common than those alleging intentional discrimination, known as disparate treatment. But in AI systems, discrimination is more likely to occur in the disparate impact context, said Littler attorney Bradford Kelley, who was chief counsel to former Republican EEOC Commissioner Keith Sonderling.

Kelley said companies should still be concerned that their algorithms could invite disparate impact complaints, noting that many states and local authorities have their own laws and regulations.

“The risk is still there,” he said.

Despite its modern-day applications, the idea of disparate impact is not new: It has been an established part of U.S. civil rights law for decades, and has helped stop unfair practices that keep minorities, women, people with disabilities, older adults, and others from getting jobs, loans, or equal pay.

The EEOC memo cites Trump’s executive order arguing that disparate impact has become a “key tool” of a “pernicious movement” that threatens meritocracy in favor of “racial balancing” in the workforce, and instructs districts to compile a list of pending disparate impact cases, then close them by Tuesday. Workers who submitted complaints solely on these grounds will receive a notice from the agency saying they can pursue the case in court on their own if they wish.

The EEOC declined to comment on the memo, citing that its contents are privileged. But a spokesperson said that as an executive branch agency, the EEOC “will fully and robustly comply with all Executive Orders” and that it “continues to fulfill its statutory obligation of accepting all charges of discrimination from all charging parties and serving all charges of discrimination on the relevant employer, regardless of the theory of discrimination alleged.”

Webber said the EEOC's decision not to pursue these investigations “definitely erects an unnecessary hurdle" for workers who filed complaints, and now no longer get the benefit of a publicly-funded investigation, which can be difficult and costly. Unlike private lawyers, the EEOC has the ability to compel employers to provide information early in the process.

Former EEOC general counsel Karla Gilbride, who was fired in January alongside two Democratic commissioners, said she was “disturbed” by the agency’s decision to close charges based on their category, not the facts and evidence.

And because disparate impact often involves company-wide or industry-wide policies, the number of people affected is usually larger than an individual discrimination complaint.

“You’re talking about thousands of people who could be affected by this,” added former EEOC commissioner Chai Feldblum.

Despite the EEOC's new stance, some businesses are proactively addressing potential bias. Job search website Indeed, for example, recently developed a virtual hiring assistant using generative AI aimed at helping employers find and connect with talent, according to senior product director Heidi Laki.

She said a core tenet of building out that product is ensuring AI is used responsibly — and conducting both disparate impact and disparate treatment analyses to “make sure that we don’t have outcomes that disproportionately affect populations."

And following the EEOC's shift in enforcement, “we are maintaining all of our responsible AI analysis that we had before,” Laki said. ________

The Associated Press’ women in the workforce and state government coverage receives financial support from Pivotal Ventures. AP is solely responsible for all content. Find AP’s standards for working with philanthropies, a list of supporters and funded coverage areas at AP.org.