For decades, the federal government has used data analysis to ferret out race and sex discrimination, winning court cases and reaching settlements in housing, education, policing and across American life. Now the Trump administration is working to unwind those same cases.
n recent weeks, the Justice Department backed out of an agreement with an Atlanta bank accused of systematically discouraging Black and Latino home buyers from applying for loans. The Education Department terminated an agreement with a South Dakota school district where Native American students were disciplined at higher rates than their White peers. And federal prosecutors have dropped several racial discrimination reform agreements involving state and local police departments — including that of Minneapolis, where George Floyd was murdered by an officer in 2020.
The Justice Department now is reviewing its entire docket and has already dismissed or terminated “many” cases that were “legally unsupportable” and a product of “weaponization” under the Biden administration, said Harmeet Dhillon, who heads the Justice Department’s Civil Rights Division.
“We will fully enforce civil rights laws in a way that satisfies the ends of justice, not politicization,” she said in a statement to The Washington Post.
The review includes cases and reform agreements forged after years-long investigations that the administration says lacked justification. Civil rights experts estimate that dozens of discrimination cases involving banks, landlords, private employers and school districts could face similar action.
“What we’re seeing is an attempt by the Trump administration to really dismantle a lot of the core tools that we use to ensure equality in the country,” said Amalea Smirniotopoulos, senior policy counsel and co-manager of the Equal Protection Initiative at the Legal Defense Fund, a nonprofit that has long advocated for the civil rights of Black Americans and other minorities.
At the center of this effort is “disparate impact analysis,” which holds that neutral policies can have discriminatory outcomes even if there was no intent to discriminate. The legal standard stems from Griggs v. Duke Power, the landmark 1971 Supreme Court decision that became a staple of civil rights litigation. In that case, attorneys relied on statistical evidence to show how standardized testing prevented Black employees in North Carolina from advancing at the energy company.
The legal theory has been consistently recognized by the Supreme Court, written into federal regulations and enshrined into employment law by Congress. But President Donald Trump declared it unconstitutional in April, issuing an executive order that kicked off an intense review of civil rights regulations, enforcement actions and settled cases.
Now, government agreements and orders that relied on disparate impact in pursuing sex, race and disability discrimination cases are being undone.
On May 23, for example, the Justice Department terminated an agreement with Patriot Bank, a Tennessee-based lender accused of failing to lend in predominantly Black and Latino neighborhoods in Memphis, from 2015 to 2020. Prosecutors used statistical evidence to show disparities in the bank’s lending practices alongside evidence of intentional discrimination, such as targeting most of its advertising in majority-White neighborhoods.
A three-year agreement to reform its lending practices had been in place for a little over a year before Trump’s Justice Department moved to end it, noting the bank was in compliance with the reform agreement. Patriot declined to comment.
Civil rights advocates worry about the future of similar enforcement.
“It’s critical we have disparate impact to fight against race discrimination,” said Thomas Silverstein, executive director of the Poverty & Race Research Action Council, who is watching at least a dozen housing and lending cases that could fall apart without it.
The principle has been fundamental to ending “practices that perpetuate inequalities in our society that have their roots in Jim Crow, redlining and the state-sanctioned discrimination of the past.”
Disparate impact has long been anathema to conservatives, who say it can result in quotas and deny equal opportunity to White people. But past Republican administrations opted not to take this issue on, partly because of Supreme Court precedent and partly because it might prove politically unpopular.
“What changed is just political will,” said Kenneth L. Marcus, who headed the Education Department’s Office for Civil Rights during both George W. Bush’s administration and Trump’s first term. “The second Trump administration is more willing to take on potentially contentious civil rights issues than any Republican administration this century.”
Trump issued a slew of executive orders to eradicate diversity, equity and inclusion, or DEI, programs — calling them “illegal and immoral” days after he returned to the White House in January — and ordered the government to close diversity offices and fire their staff.
His administration has since launched investigations into corporations, law firms and colleges over their diversity initiatives, while going to battle with Harvard University for its refusal to comply with a set of demands to alter its governance, admissions and hiring practices.
When Trump set his sights on disparate impact in April, he called it a “pernicious movement” that ignores “individual strengths, effort or achievement.” He ordered federal agencies to review any cases and reform agreements that rely on the theory — and terminate them as they see fit.
The actions are long overdue, said Dan Morenoff, executive director at the American Civil Rights Project, a nonprofit law firm that opposes the use of disparate impact and diversity initiatives. He contends that the government’s use of disparate impact has been, in many cases, legally dubious, adding that its assumptions are fundamentally flawed.
“The people who most appreciate disparate impact appear, usually, to be deeply wed to the idea that any discrepancies are best explained by discrimination,” he said. “All of us know that’s usually not true — that often different groups of people … just don’t behave the same, don’t prefer the same things, don’t pursue the same ends.”
The Supreme Court most recently upheld the use of disparate impact analysis in a 2015 housing case. But that decision was decided on a 5-4 vote in an opinion written by Justice Anthony M. Kennedy, now retired. Some conservatives believe the court’s current conservative supermajority might give them their wished-for outcome.
“It’s clear what the Trump administration is aiming for is to get this question to the Supreme Court in hopes the Supreme Court will take that tool away,” said Smirniotopoulos of the Legal Defense Fund.
In 2023, the Justice Department alleged that Atlanta-based Ameris Bank avoided providing home loans to Black and Latino home buyers in Jacksonville, Florida, in a practice known as redlining. The bank almost exclusively advertised in majority-White neighborhoods and made little effort to do business in majority Black and Latino neighborhoods, according to its lawsuit.
Only 2.7 percent of Ameris’s mortgages went to borrowers in Black and Latino communities from 2016 to 2021, the complaint said, while its competitors issued more than three times as many loans during that window. Ameris knew about the disparities but failed to correct them, the government alleged.
Though it admitted no wrongdoing, Ameris quickly settled the case, agreeing to a set of measures whose progress would be monitored by the court.
Then, on May 19, the Justice Department moved to unwind the settlement, saying that the bank has “demonstrated a commitment to remediation” while freeing it from its legal obligations to implement the reforms. The bank did not object to the move. Prosecutors did note that Ameris had disbursed the entirety of a $7.5 million loan subsidy fund for borrowers in Black and Latino neighborhoods.
A judge granted the request a day later. Ameris declined to comment.
In a Rapid City, South Dakota, case, the question centered on education of Native American students.
An investigation of the Rapid City school district by the Education Department’s Office for Civil Rights found that Native students were twice as likely as White students to be referred for discipline, more than four times as likely to be suspended and more than five times as likely to be referred to law enforcement officials.
It also found disparities in advanced coursework. At the middle school, only about 2.5 percent of students in advanced classes were Native American, while Native students made up about 18 percent of the student population. And the high school serving the majority of Native students offered four Advanced Placement courses while the other, slightly smaller, high school offered 10.
There was also some evidence of intentional discrimination. For instance, during the investigation, the then-superintendent described certain Native American tribes as not commonly valuing education and operating on “Indian time,” meaning they arrive late.
Last year, the Biden administration reached a voluntary deal with the district, which agreed to staff trainings, better communication with parents, ongoing monitoring and other steps.
“I felt a sigh of relief,” said Valeriah Big Eagle, a tribal member of the Ihanktonwan Dakota, who has three children in the district. She is one of the people who had complained to the federal authorities about what she saw as discrimination. “I was like, ‘Okay, maybe they’ll listen to the government.’ It made me feel something actually was going to be done.”
But in March, the Trump administration notified the district that it was terminating the agreement — a move the district did not request, a spokeswoman said. The government cited elements of diversity, equity and inclusion in the agreement.
It’s frightening and discouraging to Big Eagle, who hopes that district officials continue to implement changes but isn’t certain they will.
“It feels like we’re back to square one,” she said.
Meanwhile, cases involving police and fire departments have been prime targets.
In October, the Justice Department sued the Maryland state police, alleging that it discriminated against Black applicants by using a written test that was not job-related in its screening process: It tested for math, grammar, reading and reporting-writing skills. The department likewise required a physical test that screened out a disproportionate number of female applicants.
The state agreed to settle the case, committing to reforms that included finding alternatives to the written and physical tests in question. The consent order had been in place for only four months before the Justice Department backed out of the agreement.
“American communities deserve firefighters and police officers to be chosen for their skill and dedication to public safety — not to meet DEI quotas,” Attorney General Pam Bondi said of her decision in February. That day, she also announced three other dismissals involving police and fire departments accused of discrimination on similar grounds.
In May, the Justice Department moved to drop police reform agreements with Minneapolis and Louisville, with Dhillon, the civil rights division head, asserting that they relied on “faulty legal theories.”
Both cases used disparate impact analysis to allege that Black people were stopped, searched and arrested at disproportionate rates — and that race-neutral policies at both departments resulted in discriminatory policing.
