Trump’s Department of Labor claims without data that states defrauded government

Keith Sonderling, acting head of the agency, threatens to withhold administrative funds from states for first time in history

Keith Sonderling sent letters to 53 states and US territories demanding action to “combat waste, fraud, and abuse” within the unemployment insurance program, threatening to withhold administrative funds from states for the first time history.

“We are officially putting governors on notice,” said the acting US secretary of labor. “The American people will no longer tolerate the blatant waste, fraud, and abuse of their hard-earned tax dollars – no state should allow it either. If states allow it, they will suffer the consequences. This department is no longer afraid to use every lever available to ensure taxpayer money is protected.”

The agency did not provide data on fraud or alleged fraud in unemployment systems, but highlighted three Democrat-led states – California , New York, and Illinois – and made claims about each.

They say California owes $20bn to the federal government for a loan during the Covid-19 pandemic. California has struggled paying off the loan, as they did with a similar federal loan received during the 2008 economic recession, due to the current setup of how employers are taxed to fund unemployment. The unemployment payroll tax system in California has been unchanged since 1984 at a taxable wage ceiling of $7,000 on a workers’ wages and maximum tax rate of 5.4%, leaving the state with insufficient funds to cover its unemployment reserve while legislators on both sides of the political aisle have been working to try to resolve the issue.

The Department of Labor also claimed that New York loses an estimated $2m per day in unemployment insurance fraud and improper payments, but did not differentiate between the two. They also cited that Illinois has improper payments of $320m, at a rate of 14%.

Improper payments are not fraud , rather are most often cited as due to antiquated technology, with an estimated improper payment rate of 14.9% across the US.

While New York, California and Illinois have high improper payment rates compared to the rest of the US, several Republican states also have leading improper payment rates. Florida is reporting a 36.43% improper payment rate, more than double the rate in California at 16.85%, based on data from 2021 to 2024 .

“Fraud is still a problem and it hasn’t gone away since the start of the pandemic but this press release is part of the problem, why the fraud isn’t going away,” Michele Evermore, senior fellow at the National Academy of Social Insurance, told the Guardian.

She noted that states should be able to seek help from the US Department of Labor instead of being blamed solely for the issue – with threats.

“No state wants to pay fraudulent benefits to criminal actors,” she added. “It should be an all of government, all of society response, instead of calling out governors they have a political beef with. It’s not the right way to soberly and stoically deal with a problem that everybody faces.”

Sonderling appeared on Fox News on Wednesday morning to discuss the letters.

“I will essentially cut off the states’ administrative funds and then they won’t be able to administer this unemployment insurance due to the fraud,” said Sonderling, who then claimed Democrat governors are the states with the highest fraud, without citing any evidence or data to substantiate the claim.

In May 2025, the Department of Labor sent letters to state unemployment agencies requiring them to return any unspent funds allocated in the American Rescue Plan Act (ARPA), a move criticized by unemployment experts for the disruption it would cause to states in the midst of modernizing their unemployment systems and efforts to fight fraud.

Estimated fraud during the Covid-19 pandemic in unemployment insurance across the US between April 2020 to May 2023 was between $100bn to $135m , according to the Government Accountability Office, though the Department of Labor argued the estimate was overstated.