Normal view

These Local Jails Were Supposed to Detain Immigrants for Only 72 Hours. One Woman Was Held for a Total of 26 Days.

A large, modern multistory jail facility illuminated by warm evening sunlight behind a fenced parking lot.
The Pinellas County Jail in Clearwater, Florida, where ProPublica found immigrants were rebooked at much higher rates than at similar facilities across the country Tina Russell for ProPublica

Every third morning Soumia Bensalah was roused before the break of dawn. Pulled from her cell, shackled hand and foot, she was herded onto a bus and driven over Tampa Bay from the Pinellas County Jail to a cramped hold room at a U.S. Immigration and Customs Enforcement office. 

Each time, she sat there all day, still shackled. Once evening came, she was driven back to the jail, forced to wait for hours in a hallway, photographed, strip-searched, questioned and led back to her cell. This happened five times last December, and Bensalah said it left her exhausted and humiliated.

It was no coincidence that Bensalah, a French Moroccan with a pending green card application, was released from the Pinellas jail and then sent back at regular, 72-hour intervals, according to legal experts. That’s the maximum amount of time that county jails like Pinellas’ can hold immigrants on behalf of ICE. The limit exists to prevent immigrants detained on civil charges from spending too long in jails meant for people detained on criminal charges. 

Repeat releases and readmissions, known as rebookings, typically happen in secret and circumvent federal rules and regulations, according to eight legal experts interviewed by ProPublica. 

The jail had refused to release the logs that documented Bensalah’s journey, citing federal immigration regulations that shield jail records, which are typically public. We were able to trace the journey after obtaining an unredacted copy of the Pinellas jail’s booking logs from a source who provided them on the condition of anonymity, for fear of workplace reprisal. We checked them against anonymized federal immigration data collected by the Deportation Data Project.

The repeat rebookings stem from the fact that the federal government often doesn’t have enough detention space to jail all the immigrants whom ICE and its cooperating local authorities are rounding up. The shortage of detention space is pronounced in Florida, where the majority of law enforcement agencies have signed agreements to collaborate with ICE on immigration enforcement. 

ProPublica’s analysis of data obtained by the Deportation Data Project shows that over the past year, jails in both Pinellas County and in Orange County, Florida, had unusually high numbers of repeat rebookings of immigrants held for ICE compared with other jails nationwide. (Read more about our methodology here.) The Orange County Jail, in Orlando, rebooked 559 of them two or more times from July 2025 until early February 2026, when the data shows the practice ceased. Less than two hours away, in the Tampa Bay area, Pinellas rebooked 174 immigrants at least twice from July 2025 until the practice appeared to cease in July 2026, when zero rebookings were recorded in the data. 

The scale in the two Florida counties is unmatched, ProPublica found. Other jails across the country rebooked no more than 10 immigrants two or more times.

All of the rebooked immigrants ProPublica counted were jailed for ICE, meaning they were not being detained on local charges. Our analysis counted a person as having been rebooked only if they cumulatively spent at least 72 hours in jail.  

Emily Ryo, a Duke University law professor who studies ICE detention, said jails often can’t provide comprehensive medical screenings, space for confidential legal consultations or even outdoor access. All of those things should be guaranteed to immigrants in ICE custody, according to ICE’s own detention standards. 

Two former Department of Homeland Security officials who spoke with ProPublica described the repeated rebookings as an attempt to evade the detention standards.

“What it shows is that they were very aware that they were breaking a rule,” said Michelle Brané, a former DHS immigration detention ombudsperson. 

Bensalah landed in jail after her U.S. citizen husband allegedly assaulted her, according to a police report. She called the police, who observed cuts on both and booked them into the Polk County Jail on battery charges. (The charges against both were later dropped.) She posted bond but was not released because of an ICE detainer, a written request from the federal government asking a jail to hold someone for up to 48 hours after they’d otherwise be released. A detainer is meant to give ICE time to pick up an immigrant and begin deportation proceedings. 

State law requires every jail in Florida to sign collaboration agreements with ICE. Most jails in the state can detain immigrants for ICE for only up to 48 hours. Through June, Pinellas and Orange had special contracts that ICE says allowed them to hold immigrants for up to three days. ICE designates them as “under 72” facilities in its database.

“At 72 you should have walked him out the door,” said a former senior DHS official, who spoke on the condition of anonymity because of fear of reprisal. 

Instead of sending Bensalah to a longer-term immigration detention facility in the days after she posted bond, ICE moved her two counties over to Pinellas. Three days later, her rebookings began. 

“There’s no argument that this is a lawful detention,” Jennifer Chacón, a professor of law at Stanford University, said of the repeat rebookings.

In response to ProPublica’s questions about the repeat rebookings, a DHS spokesperson wrote: “ICE is NOT breaking any laws by making custody determinations based on bed space and ensuring detainees’ presence for immigration proceedings or removal from the United States.”

ProPublica shared with DHS details about Bensalah’s case. The department responded that “she received due process.”

“Under President Trump and Secretary [Markwayne] Mullin, if you break the law, you will face the consequences,” DHS wrote. “Criminal illegal aliens are not welcome in the U.S.”

All the people rebooked at least twice in Pinellas and Orange counties ended up behind bars for more than 72 hours. One woman from Mexico was rebooked into Pinellas 10 times, spending a cumulative 26 days in the jail. (DHS declined to comment on her case because ProPublica did not provide her name; she had declined multiple requests for an interview.) She was then transferred to an ICE facility and eventually released after posting a $2,000 bond, according to Deportation Data Project figures. Only 65 people rebooked into those counties’ jails two or more times through June 2026 were eventually released on bond.

Of the 423 immigrants rebooked at least twice in Pinellas and Orange counties through December 2025, 115 — or 27% — had no criminal charges or convictions except for traffic-related offenses. That is consistent with recent data released showing that an increasing percentage of all ICE detentions involve people with no criminal record or charges.

10 Rebookings at the Pinellas County Jail

In late 2025, federal and state officials kept a Mexican woman in custody for almost a month by repeatedly rebooking her just before her 72-hour time limit ran out.

A timeline showing a Mexican woman’s 10 rebookings, each within 72 hours, during a nearly monthlong stint, from Sept. 23 to Oct. 22, at the Pinellas County Jail in Clearwater, Florida.
Note: ProPublica withheld the identity of this individual after she did not respond to multiple requests for comment. Source: Deportation Data Project. Chris Alcantara/ProPublica

For some rebooked immigrants in Orange County, the turnaround time was extreme: 30 of them were rebooked in 15 minutes or less, right as their current stay approached 72 hours. Those rebookings were between September 2025 and February 2026. Orange County rebooked five immigrants back into the jail in as little as one minute. 

Additionally, the Pinellas County Jail has held at least 157 immigrants for ICE for longer than three business days in a single booking since the start of President Donald Trump’s second term in January 2025, according to a ProPublica analysis. In the two years prior, it held only five immigrants for longer than three days. After one woman — a Brazilian immigrant who had not been charged with a crime — challenged her nine-day detention in federal court this July, ICE released her, calling her jail stay a “procedural error.” 

“All her claims will be heard by an immigration judge, and she will receive due process,” DHS wrote in response to questions about the Brazilian woman’s case. “This administration will not ignore the rule of law.”

A secured jail entrance featuring a guard booth, red traffic barricades, orange traffic cones and green directional road signs.
ProPublica found that all of the people rebooked two or more times in Pinellas County Jail ended up being held for more than 72 hours. One woman from Mexico was rebooked 10 times, for a total of 26 days in county detention. Tina Russell for ProPublica

Legal experts say that the only real recourse for immigrants repeatedly rebooked or held for prolonged periods in a local jail is to file an emergency petition in federal court seeking relief from illegal detainment. ProPublica was able to determine that at least 29 of the immigrants held for ICE in Orange or Pinellas counties filed such petitions, called habeas corpus claims.  

In response to the Brazilian woman’s habeas petition, the Pinellas County Sheriff’s Office claimed that she was in federal custody and that it didn’t have the power to release her. Federal appeals courts have found the opposite: Jails have the power to release ICE detainees after a hold expires and can be found liable if they don’t. Last November, a federal jury ordered Suffolk County, New York, to pay $112 million to more than 600 immigrants it found had been wrongly held. Suffolk County has appealed the decision. 

Corporal Jamie Miller, a spokesperson for the Pinellas County Sheriff’s Office, told ProPublica that the facility holds people for ICE for only up to 72 hours before releasing them. The sheriff’s office declined further comment after ProPublica shared the findings from its analysis and sent a list of questions. The office declined requests for an interview with the sheriff.

In a written statement to ProPublica, a spokesperson for the Orange County Jail said it had eliminated “multiple bookings.” The statement referred further questions to ICE. 

Our analysis shows that in Orange County, the rebooking practice stopped in early February. That change happened after widespread public opposition. In Pinellas, where opposition has been quieter, the rebookings slowed in January, but single stays exceeding 72 hours started to climb. 

15 Rebookings at the Orange County Jail

Officials rebooked a Brazilian woman more than a dozen times during a nearly two-month stint.

A timeline showing a Brazilian woman’s 15 rebookings during a seven-week stay, from Oct. 31 to Dec. 19, at the Orange County Jail, in Orlando, Florida.
Note: Because federal immigration data is anonymized and the Orange County Jail declined to produce its booking logs in response to a records request, ProPublica could not identify this individual. Stints that were longer than 72 hours did not qualify as overstays for this individual because each included weekends, which do not count, according to the jail’s then-contract with federal authorities. Source: Deportation Data Project. Chris Alcantara/ProPublica

Orange and Pinellas counties both stand out nationally for their volume of rebookings. But their differing politics contributed to different outcomes at their jails. 

In majority-Democratic Orange County, religious groups, attorneys and a coalition of activists prompted officials and the courts to take action. In county commission meetings, community members told the stories of immigrants held in the jail and pointed out that the county was losing money every time it held someone for ICE.

County officials — caught between a state legislative mandate to cooperate with ICE, a national mass deportation push that was filling their jail with detainees and a vocal electorate that demanded change — wrestled with how to respond.

In August 2025, the Orlando Sentinel reported on half a dozen rebookings in the Orange County Jail. Months later, in late January, a defense attorney showed up at a county commission hearing to describe how the rebooking practice was still going on. She claimed she had multiple clients who’d been illegally held at the jail on behalf of ICE and that the rebookings violated due process.

A week later, County Mayor Jerry Demings sent a letter to ICE, informing it that “ICE inmates will be limited to one continuous housing period of up to 72 hours per immigration matter.” After the letter, the rebookings ceased, ProPublica’s analysis shows.

The day Demings sent the letter, Orlando federal judge Paul Byron criticized the rebooking practice at a hearing for an immigrant who’d claimed he was illegally detained.

“If the government is pulling a detainee out of jail to artificially reset the clock, that is wildly inappropriate,” Byron told Assistant U.S. Attorney Phillip Ragler. “What allows that?”

Ragler didn’t have an answer. Byron ordered the immigrant be freed.

Ragler recently told ProPublica that the experience led to a crisis of conscience.

“Being forced to defend ICE’s actions in a professional capacity caused significant moral and ethical conflicts for me personally,” Ragler wrote in a statement to ProPublica, adding that ICE gave the U.S. attorneys no reasoning or justification for detaining people, many of whom had no criminal record. He told ProPublica he resigned from the position shortly after the hearing, following almost a decade in the office.

Ultimately, Orange County renegotiated its contract with ICE so that immigrants could be held for only two days. Soon after, the number of immigrants in ICE custody booked into the jail plummeted, ProPublica’s analysis shows. Just two were booked into the jail in July, down from a high of 692 in January.

At around the same time that Orange County stopped repeat rebookings of immigrants, the number of rebookings in Pinellas started decreasing. But unlike Orange County, Pinellas then began holding more immigrants for longer than three business days in a single stay — and saw no dip in the number of immigrants it was booking overall. 

Six months later, in July, Pinellas appears to have stopped repeat rebookings of immigrants, ProPublica’s analysis found. That same month, at least 12 immigrants booked in the jail were held for more than three business days. One Honduran man stayed for at least eight days.

A white transport van with barred windows driving on a multi-lane highway near the coast.
A private prison transport truck traveling toward the ICE detention center near Tampa International Airport Courtney Prokopas

Pinellas County, unlike Orange County, trended Republican in the last presidential election. Pinellas Sheriff Bob Gualtieri has long advocated for his and other local jails to collaborate with the federal government on immigration enforcement. Over the past decade, he helped pioneer a type of contract with ICE that shields local jails from legal liability when they hold someone for the agency. He also helped create a federal-local agreement that lets jail staff serve ICE detainers on people held at the facility.

“For years, sheriffs have had to choose between releasing criminal illegal aliens from their jails back into the community, or exposing themselves to potential civil liability,” Gualtieri is quoted as saying in a 2017 ICE press release.

In the August 2025 Orlando Sentinel article, Gualtieri is quoted as saying that there’d been rebookings at his facility due to “a capacity issue” at the federal detention facility where they were supposed to be transferred. 

In the summer of 2025, Gualtieri was part of a group of Florida sheriffs that proposed allowing ICE to take advantage of unused jail beds across Florida — and have Florida police transport the detainees. Since Trump took office in January 2025, Gualtieri’s own jail began to house a historically high number of them. 

More than 7,000 immigrants have been detained for ICE in the Pinellas County Jail during the first 564 days of Trump’s second term, according to ProPublica’s analysis. That’s more than four times as many as during the last two years of the Biden administration.

The jail’s proportion of ICE detainees among its total bookings steadily increased over the course of 2025, from under 7% at the start of the year to 18% in December, according to ProPublica’s analysis of Pinellas jail booking data. In 2026, the proportion hovered around 20% from January through July.  

Most of the immigrants ProPublica tracked in Pinellas and across the country, including Bensalah, ultimately were deported. After her fifth rebooking, she was transferred to the first of four ICE detention centers, then deported to France. 

Paul Chavez, litigation director for Americans for Immigrant Justice, a legal nonprofit that litigates for immigrants’ rights, said repeat rebookings like Bensalah’s are clear examples of unconstitutional detention. 

“If it was that easy to thwart, all we would have to do to thwart the Fourth Amendment was throw somebody in the backseat of a car, drive them to McDonald’s and then drive back,” he said. “I don’t know that the drafters of the Fourth Amendment would have found solace in that.” 

How We Identified People Rebooked Into County Jails

ProPublica found immigrants ICE repeatedly rebooked into county jails by analyzing multiple federal and local detention databases. 

To count immigrants rebooked two or more times at jails nationwide, we used federal immigration detention data obtained by the Deportation Data Project. The data assigns a unique ID to each person in detention and records the date and time each person is booked into and out of each facility. We calculated the length of each stay, as well as the period between leaving and returning to a single facility.

A detainee may leave ICE detention, such as when they are released on immigration bond or into U.S. Marshals’ custody and then return to ICE custody weeks or even months later. To avoid counting these cases, ProPublica counted a person as being rebooked only if their return to the same facility occurred within 24 hours. Detainees are sometimes rebooked into the same facility after longer periods, but our analysis found that 95% of detainees were rebooked within 12 hours.

The detention data covers a period from Oct. 1, 2022, through Aug. 6, 2026. We limited our analysis to local jails, where federal rules require jails hold ICE detainees for less than 72 hours unless jails sign a special contract with ICE and meet ICE’s detention standards, which include submitting to regular federal inspections. Because a single rebooking could be due to a failed transfer or a missed or delayed deportation flight, we counted only immigrants who had been rebooked two or more times. An expert familiar with ICE detention practices told ProPublica that was a better indicator of intentional rebookings made to extend detentions in a single jail. 

Under federal regulations, weekends and holidays do not count toward a jail’s stay limit. ProPublica excluded these days when counting the number of hours each person spent in a county jail. We counted someone as having been rebooked only if their total cumulative stay reached at least 72 hours. When tallying the total stay of individual detainees, ProPublica counted all hours spent in jail, including weekends and holidays.

To count the number of rebooked detainees without criminal charges or convictions, ProPublica used ICE detention data released by the federal government under the Freedom of Information Act. Versions of this dataset were originally released to the Transactional Records Access Clearinghouse and The New York Times. This data ends in mid-December 2025. To determine that 27% of detainees rebooked two or more times had no criminal charges or convictions except for traffic-related offenses, we filtered by detainees charged with only traffic-related offenses or who were described as “other immigration violator,” a category used by the Department of Homeland Security when a detained individual has no other record of criminal charges or convictions. 

ProPublica queried an unredacted copy of the Pinellas County Jail’s inmate search database it obtained to calculate the share of bookings at the jail that were ICE detainees. ProPublica verified the veracity of the unredacted database by comparing its booking records to those in Pinellas’ online search tool and by validating the booking dates and timestamps against the federal immigration datasets.

The post These Local Jails Were Supposed to Detain Immigrants for Only 72 Hours. One Woman Was Held for a Total of 26 Days. appeared first on ProPublica.

How to Check In on Your 401(k) Fees

Illustration of a machine X-raying a piggy bank lying on a table. The scan shows dollar bills, coins and a Bitcoin symbol. Doctors stand outside the room and look in.
Tomi Um for ProPublica

The rules governing your retirement plan may be changing.

As we reported in July, President Donald Trump wants to encourage 401(k) plans to invest more in private equity, real estate and cryptocurrency, which can be complex and risky. To pave the way, the Department of Labor has proposed rules that would make it harder for employees to hold companies liable for how they oversee retirement plans.

Employers are required to serve the best interests of their employees when choosing investment options. Under the proposed changes, a company that follows a certain process would get the benefit of the doubt in court, making it harder for workers to sue over the management of their 401(k).

A spokesperson for the Department of Labor did not respond to requests for comment. In an announcement from March, the department said the proposed rule change will “democratize access to alternative investments in 401(k) plans” and “lower litigation risks” for employers who are acting with good judgment.

The new rules, expected to be finalized this year, also raise broader questions: What investment options are already offered in people’s retirement accounts, and what are people paying in fees?

We’ve asked the public to send us their annual 401(k) disclosures so we can understand the range of funds and fees people have in their accounts. (Want to send us yours? Here’s how.) So far, we’ve heard from more than 200 people, many of whom also asked us questions about their plans. To find the answers, we spoke with eight economists, lawyers and retirement benefit experts.

If you have questions about your specific retirement plan, experts recommend talking to a fee-only financial planner who will serve as a fiduciary (which means they must provide advice in your best financial interest, not theirs). You can check a financial adviser’s qualifications by searching here.

Table of Contents

How could the proposed changes affect my 401(k)?
How do I know if my 401(k) is in good shape?
How can I tell if I’m paying high fees?
How can I compare my 401(k) to plans offered by similar companies?
What if I don’t like my retirement plan?

How could the proposed changes affect my 401(k)?

Under the Labor Department’s proposed rule change, employers who show that they’ve considered a set of six factors when choosing investments and document their reasoning when building a plan should have greater protection from a potential lawsuit.

That liability tends to sit with your employer because they have final say over your plan’s investment options, even though companies usually hire firms like Fidelity or Vanguard to run the plan, and those firms often suggest which funds to include.

A company could document its reasoning for choosing to offer certain investment options and still act imprudently, said Tim Hauser, who was the deputy assistant secretary at the Labor Department’s Employee Benefits Security Administration until last December. In his three-plus decades there, he encountered cases in which companies “generated a lot of paper” to explain their reasoning but, ultimately, didn’t make wise financial decisions based on the information available to them.

Have a 401(k)? Help ProPublica Investigate What’s Really Happening to Your Money.

Some retirement plans are stuck in investments with bloated fees and costly add-ons. Share your standard plan documents with us to help reveal which financial products companies are pushing, and what they’re charging workers.

Monique Morrissey, a senior economist for the Economic Policy Institute, wrote in a June letter that the proposal would “gut protections for retirement savers” and prioritize maximizing investment returns over balancing risk. She cited a 2025 AARP survey finding most Americans don’t think it’s important to be able to access private market investments or cryptocurrency in their retirement accounts.

But Bonnie Treichel, the founder of Endeavor Retirement, a consulting firm for retirement advisers, said the proposed rules are a framework not a mandate. Employers could offer these investments, but that doesn’t mean they’ll rush to add riskier investment options.

To find out if the rule change may affect your 401(k), ask your plan administrator, often someone in your human resources department, whether the company plans to offer new investment options. (The plan administrator is often listed on the same document that lists your fees; find out how to access that information in the form on this post.)

How do I know if my 401(k) is in good shape?

For many people, a 401(k) only needs close attention when starting a job, when leaving a job and when closing in on retirement. The rest of the time, it mostly runs itself. Still, experts we interviewed said it’s worth checking in on the plan once a year as fees and fund options can change.

A healthy account comes down to three basics: how much you save, how much risk you take and how much you pay in fees, experts say. Only the first is entirely up to you. The other two you manage from a menu of options your employer builds.

The more you save — and the earlier you start — the more time your money has to grow. Companies often offer to match part of your contribution, so if you can swing it, experts advise putting in at least enough to collect the full match. Otherwise you’re leaving money on the table.

You also control what your savings are invested in, based on the menu of investment options your plan offers.

Each fund you invest in carries a different level of risk. Stocks are more volatile than bonds, for example, and a balanced investment portfolio weighs the level of risk depending on how far you are from retirement. Younger investors decades away from tapping into their retirement fund may invest more in stocks, while older workers may prefer to have more in bonds.

The “easy button” is to choose a target-date fund that’ll automatically reallocate your investments to reduce your risk by shifting from stocks to bonds, or other less volatile investments, as you approach retirement.

“For most people, this is all the money they have,” said Jean-Pierre Aubry, an associate director at the Boston College Center for Retirement Research. “It’s not money you want to play around with.”

How can I tell if I’m paying high fees?

Compounded over time, even small differences in fees have a huge impact on your retirement savings. By the Labor Department’s own math, 1% in additional fees can shrink a nest egg at retirement by 28%.

Federal law requires employers to make sure employees are paying reasonable fees. Since 2012, the Labor Department also requires companies to send a disclosure form to employees listing all funds and fees. (That’s the annual disclosure we’re asking people to send to us. If you want to send us yours, find instructions on the form in this post.) Still, the Government Accountability Office found nearly 4 in 10 people don’t fully understand the fees they’re paying on their retirement plan.

The number you should watch for is each fund’s expense ratio: the fees firms charge employees for owning or investing in a fund, taken as a percentage of the money you have in it. An expense ratio of 0.5% means you pay a $5 fee per year for every $1,000 you’ve invested. To find the fees you’re paying, look for your annual disclosure on your plan’s website or ask your HR department. (See instructions for your specific plan here.)

If most funds on your 401(k) menu have an expense ratio over 1%, that should be a “red flag” that it’s a high-cost plan, said Christine Benz, director of personal finance and retirement planning at Morningstar.

Even 0.75% or 0.5% is “actually pretty high by 401(k) standards,” said Quinn Curtis, a law professor at University of Virginia who has studied the litigation around 401(k) fees.

The cheapest options are typically index funds, which mirror a slice of the market like the S&P 500 instead of paying a manager to pick stocks. Workers are gravitating to these passive, low-cost investments and the firms are competing to win over investors, which is driving down fees even more. Most index fund fees are under or around 0.1%, Benz said. If your index fund is charging several times that, experts say it’s worth asking why or finding another option.

When you check your 401(k) plan once a year, make sure you’re still in the funds with the lowest available fees, Aubry said.

And be wary of offers for personalized portfolios at the click of a button, Treichel said. These services often charge additional fees.

“Nothing in life is free,” she said. “So, if it looks free, look further.”

How can I compare my 401(k) to plans offered by similar companies?

There’s no easy way. No free database exists where you can, say, type in your employer name and see how its 401(k) investment options compare to what other companies offer. But there are a few general trends to be aware of.

The size of your company matters. Smaller companies tend to get offered retirement plans with higher fees, according to research. Larger companies have more employee money and therefore leverage, and tend to negotiate lower fees. So, if you work for a smaller company, experts said you may want to be especially vigilant about your plan’s investment options.

Other signs of a good plan: your employer pays some or all of the administrative costs and your employer offers a matching contribution, experts said.

And more choices don’t necessarily mean better ones. Most people don’t have the time or the expertise to understand the investment funds available to them, Curtis said. A well-designed retirement plan gives workers a curated set of low-cost investment options that’ll “meet the needs of most investors.”

What if I don’t like my retirement plan?

Start with your plan administrator, often someone in your company’s HR department, experts say. Tell them that you aren’t happy with the investment offerings and the associated fees, and lobby for additional options.

If you believe the people overseeing your 401(k) plan have been careless, disloyal or negligent, resulting in a smaller nest egg for your retirement, you can contact the Department of Labor, Hauser said. An employee benefit adviser should follow up with you to learn more and may refer your case to the enforcement division for further investigation.

Under federal law, you can also file a lawsuit against your employer.

But Hauser said it may be difficult for the average person to police the options in their 401(k) plan. Much of the responsibility falls on employees, whether or not they are well-versed in investing, he said. That can mean people invest in their retirement fund for decades before they notice the high fees that cut into their potential retirement savings.

“People consistently underestimate the impact of even small differences in fees,” Hauser said.

The post How to Check In on Your 401(k) Fees appeared first on ProPublica.

The Costly Blunders in the Army’s $533 Million Fiasco

A man in an Army combat uniform speaks to a group of defense officials in a factory, in front of artillery shells at various stages of production.
Former Army Secretary Christine Wormuth joined defense officials for a facility tour at the opening of General Dynamics’ artillery factory in 2024. Sgt. 1st Class Nicole Mejia/U.S. Army

The U.S. Army paid General Dynamics $533 million for an artillery factory that failed to produce a single usable shell.

Instead of efficient, state-of-the-art production lines, half a billion dollars in taxpayer funding paid for machines that kept failing in bizarre ways, former workers told ProPublica’s Jesse Coburn. Giant robot arms would catch fire. The Texas factory’s signature device often cracked the steel meant for the shells. And workers regularly had to take a sledgehammer to certain machines to try to make them function properly. But the machines still botched nearly every shell.

The boondoggle, which included a real-life dumpster fire, was described as an “absolute disaster,” by one former official who worked in an Army office overseeing the project. The official, like others interviewed for the story, spoke on the condition of anonymity.

In a detailed statement, the Army told ProPublica it exercises “rigorous oversight” and that, “where vendors fail to meet contract specifications” — as at the General Dynamics artillery factory — ”we are evaluating contract performance, seeking recoupment of funds, and pivoting resources.”

General Dynamics declined an interview request, but the company has said it “met or exceeded requirements,” according to a Department of Defense inspector general report. In a statement, a company spokesperson told ProPublica its reporting “fundamentally mischaracterizes the circumstances,” but did not respond to a request to specify our reporting’s alleged errors.

To chronicle what went wrong, Coburn interviewed 36 people who’ve worked for the Army, the Pentagon, General Dynamics and the White House, and he reviewed internal company documents as well as photos and video from inside the factory. Here are the lapses in planning and execution Coburn’s reporting uncovered — and some of the consequences of those mistakes.

The Red Flags

The Army awarded the project to General Dynamics in a rushed process despite significant unknowns.

When Russia invaded Ukraine in 2022, the Biden administration rushed to boost production of artillery shells to support Ukraine’s war effort. General Dynamics was the only company producing the 155 mm metal shell bodies in the U.S. at that time, mainly in a century-old facility in Pennsylvania.

The company could have simply replicated the facility’s traditional manufacturing method, which dated back to the time of the Korean War, to ramp up production. But General Dynamics proposed using a state-of-the-art production line from Repkon, a Turkish company virtually unknown in American defense. 

Repkon said other countries were already using its machinery to build an older, simpler model of the artillery shells. But it wasn’t clear the Turkish equipment could work with the particular steel used to make the Army’s newer model of 155 mm shells. And Repkon said the Army and General Dynamics could not inspect full artillery production lines in action, citing customer privacy.

Repkon did not respond to requests for comment. 

The Department of Defense did not competitively bid the project.

The Army could’ve insisted that General Dynamics more fully demonstrate that Repkon’s machines could perform the desired work. But that would’ve taken time, and — as a former Army official told Coburn — “there was incredible pressure to go fast.”

In November 2022, the Army gave General Dynamics the first in a series of contract awards for the artillery factory. Then, Congress granted the Department of Defense the power to award money for Ukraine-related causes without some of the usual contracting safeguards meant to ensure taxpayer money doesn’t go to waste. That meant the Army could give General Dynamics no-bid awards and the company could start work on the project before finalizing the contract terms with the government.

The U.S. ended up ordering three Repkon production lines — instead of just one — without knowing for certain whether they could make shells that met the Army’s specifications. (The Army said it selected General Dynamics because of the company’s unique artillery production experience).

The Army did not thoroughly vet the proposal before approving it.

Both the Army and General Dynamics sent staffers to Turkey to inspect the Repkon machines. But they never inspected a full production line in action, and they didn’t send enough experts who could identify potential defects in Repkon’s equipment, four former General Dynamics and Army officials told ProPublica.

The Army also didn’t require General Dynamics to demonstrate it could use Repkon’s equipment to complete the entire production process and make shells that met the service’s specifications. (The Army said that “formal testing could not occur prior to full machinery installation.”)

In 2024, leaders from the Army and General Dynamics gathered in Texas to celebrate the factory’s opening. But the machines were barely functioning. 

Here’s how the rushed planning from the Army and General Dynamics played out in the factory.

The Consequences

In Texas, robotic arms caught fire and smashed into equipment.

Twelve former factory workers told Coburn that problems quickly piled up inside the factory. Machines meant to begin giving the shells perfectly smooth noses would, instead, mangle them into swirls that looked like soft-serve ice cream.

The robotic arms had a tendency to swing out of control, smashing into things around the factory. Workers talked about the arms going “rogue.” Sometimes, a machine in the factory would be seen moving on its own, controlled remotely by someone in Turkey, spooking workers in the Texas factory. 

Workers said smoke hung in the air inside the factory, and it’d reach desert-like temperatures when the furnaces or forging presses were on. Making matters worse, walls cracked, water flooded in when it rained and employees said the foundation of the buildings appeared to be shifting or sinking beneath them.

As progress at the artillery plant continued to stall, paranoia grew and rumors swirled among the American workers that the Turkish employees from Repkon might be purposefully sabotaging the machinery.

“It was astonishing how little progress we would make, month after month, year after year,” one former worker told ProPublica. “People started speculating, ‘I wonder if they’re spying on us.’” (The Army said it has no evidence of sabotage or spying.)

General Dynamics failed to meet a number of milestones at the facility, including failing to perform scheduled first article tests, which would’ve demonstrated the factory could produce shells that met the Army’s requirements. 

The Army hasn’t required General Dynamics to pay the money back.

General Dynamics, one of the world’s largest defense contractors, and Repkon haven’t been held publicly accountable for the taxpayer-funded failure. The Army also hasn’t made General Dynamics pay back the money. (The service said it will recover funding from the project by getting unspecified discounts from General Dynamics on production orders.)

In August 2025, the Army halted work on two of the three production lines at the factory. But the Army didn’t unilaterally terminate the contract awards. In December 2025, the government paid the company $26.3 million in “progress payments” for two production lines — although the lines had never made a usable shell.

The Army says it won’t spend any more money on the artillery plant. But since the Army partially halted work at the factory last summer, the same unit of General Dynamics responsible for the struggling facility has won contract awards worth $2.5 billion, according to the Army. (The service said this was for “distinct production lines and critical national defense requirements” unrelated to the Texas factory.)

Having experienced the failure of a little-proven technology in its factory, General Dynamics has announced it will team up with yet another unheard-of partner promising technological innovation. The new solution? Artificial intelligence.

Read our full investigation here.

The post The Costly Blunders in the Army’s $533 Million Fiasco appeared first on ProPublica.

Idaho Law Protects Faith Healers. It Wasn’t Always That Way.

A gravestone with various statues: two frogs smiling, a cherub praying and a childlike angel. Brown grass stretches out behind the gravestone and mountains and a blue sky can be seen in the distance.
Members of the faith-healing sect Followers of Christ bury loved ones, many of them children and infants, in Peaceful Valley Cemetery in Caldwell, Idaho Sarah A. Miller for ProPublica

In Idaho, parents who say their faith prohibits lifesaving medical care for their children can’t be charged with child neglect or manslaughter. Supporters of this protection over the years have included lawmakers who voiced support for “traditional” Idaho values.

But from at least the 1880s through the early 1970s, it was illegal for parents in Idaho to choose prayer over medical care when a child’s life was at stake, newspaper archives and historical documents show.

“Every parent of any child who willfully omits, without lawful excuse, to furnish necessary food, clothing, shelter, or medical attendance for such child, is guilty of a misdemeanor,” reads the 1887 edition of Idaho territory statutes, published before statehood.

The statutes included no mention of spiritual belief as a “lawful excuse.” Three years later, the state constitution explicitly put limits on Idaho’s freedom of religion. It would not excuse “acts of licentiousness,” polygamy or other practices “inconsistent with morality or the peace or safety of the state.”

It wasn’t long before the law came down on members of the Followers of Christ, a faith-healing sect that arrived in Idaho in 1899 and remains active in the state. From 2015 to 2025, the sect’s congregations in Idaho experienced 15 preventable deaths of children, according to ProPublica’s reporting.

Pearl Annis was a 13-year-old girl whose parents had moved from Oklahoma to Idaho’s Magic Valley in the spring of 1915 with at least one other Followers family. The Annises and their 13 children shared “a two-room shack,” a local newspaper reported. An officer and a doctor visited the home after neighbors voiced concern. They found Pearl in bed, fully dressed and on the cusp of death.

Her father, Lurid P. “Lewis” Annis, was arrested on a misdemeanor charge of refusing to provide medical attention. Pearl was taken to the hospital, according to the news story, and died there with what today would be described as septic shock from a bowel blockage. The newspaper, which didn’t appear to follow up on Annis’ criminal charge, called the Followers church “a religious cult” and Pearl’s death a consequence of “religious mania.”

News archives and historic records don’t say how many times Idaho acted on similar cases in the decades after.

But one legal battle over lifesaving care drew attention in the 1960s, as sickness made its way through a Followers family in a Boise suburb. The mother died in December 1965. Her 4-year-old daughter died the following June.

Days after the girl’s death, a judge ordered the state to take custody of her 10-year-old brother under Idaho’s child protection law. The boy was admitted to a local hospital with pneumonia, a complication of the measles. He spent days in critical condition but survived. The judge told the father that to regain custody, he would have to be willing to give the boy medical care as the law required.

A doctor testified that without the hospitalization, the child “almost certainly would have died,” a newspaper report said.

The father’s lawyer told the judge that his sect was rooted in “faith that God will heal the sick.” The law in the 1960s was clear, though: a parent had a duty to protect their child’s health, no exceptions. The judge told Samuel’s father: “The law governs conduct. Religion governs opinion.”

But the legal landscape was about to shift. It came with little public attention.

During a full-on rewrite of Idaho’s criminal code in 1971, the Legislature added a section that said the state could not bring endangerment charges against someone “who chooses for his child treatment by prayer or spiritual means alone.” Legislative records don’t make clear how the wording originated. The passage of another law one year later, affirming the “spiritual means” language and applying it to other crimes, has been credited to the Christian Science church, another Christian denomination with Idaho members.

(The Christian Science website says the church’s practice of “healing in the way Christ Jesus taught” is not “faith healing” and that members are free to choose any form of healthcare. The church also supported Oregon’s full removal of its faith-healing exemption in 2011.)

Faith-healing exemptions soon became widespread nationally. The trigger was the federal Child Abuse Prevention and Treatment Act of 1974, which set out to standardize the nation’s child welfare system by giving states grants to more effectively investigate reports of child abuse.

The Department of Health, Education and Welfare interpreted the law as requiring states to have a faith-healing exemption to qualify for the grants — an interpretation that records from the time don’t explain.

Most states went along, and Idaho further broadened its exemption in 1976 to not only keep faith healers from going to jail if their children died, but also to keep their children from being placed in foster care solely for lack of medical care.

At the national level, the mandate lasted less than a decade. The Reagan administration adopted new regulations in 1983 that took a neutral stance on the subject, and several states in subsequent years rolled back their exemptions.

Colorado, for example, reverted to prosecuting faith healers as it would any other parent in 2001 following deaths in the Followers-adjacent sect Church of the First Born, including the death of a 13-year-old girl from untreated diabetes.

But Idaho clung to its special treatment for faith healers.

It has remained that way ever since.

The post Idaho Law Protects Faith Healers. It Wasn’t Always That Way. appeared first on ProPublica.

Top Commanders Told Hegseth Not to Cut Civilian Protection Program, Poll Showed. He Did It Anyway.

A man in a dark suit sits in the front row of an auditorium full of people in military uniforms.
Defense Secretary Pete Hegseth, front left, and senior U.S. military leaders listen to President Donald Trump speak at Marine Corps Base Quantico in Virginia in 2025. Evan Vucci/AP Photo

Shortly after taking office last year, Defense Secretary Pete Hegseth polled the nation’s top commanders about cutting a program designed to protect civilians during U.S. military operations.

Hegseth had bet senior officers would endorse his view that focusing on civilian security was “woke” and out of step with “maximum lethality,” current and former defense officials told ProPublica.

Instead, every commander rebuffed the idea.

The poll from early 2025, which has not been previously reported, cemented top-level military support for Civilian Harm Mitigation and Response, a relatively new program adopted after years of heavy civilian tolls in Iraq, Afghanistan and elsewhere. As part of its mission, CHMR (pronounced “chimmer”) stationed personnel within major regional commands. Officials tried to develop a picture of life on the ground in combat zones to aid decision-making on military strikes and avoid or minimize civilian harm.

In the poll, Hegseth gave commanders three options: cut 65% of staff but keep the effort operational, continue with only a small centralized office in Washington, or eliminate the program.

Nine commanders chose the first option to preserve the enterprise as much as possible, according to a memo reviewed by ProPublica. Leaders of two commands, one overseeing the nuclear arsenal and the other focused on cyber threats, declined all three options, asking to maintain the full program.

Leaders at U.S. Central Command, the Tampa, Florida-based center that oversees military action in the Middle East, pressed to keep its full CHMR staff of 16. Eliminating or scaling back the effort would “result in decreased efficiency of U.S. lethal operations due to an increased risk of target misidentification and reduced battlespace awareness,” a memo reviewed by ProPublica said.

Retired Army Gen. Erik Kurilla, who was then in charge of Central Command, and several other senior officers directly involved in combat operations had appealed to Hegseth to keep the program. They explained that it was a valuable tool, another retired general said — but “that intent was overridden.”

Hegseth gutted the effort, leaving only a handful of people to carry out the mission. It was stripped of almost all of its funding, even as the U.S. launched new campaigns in Somalia, Yemen and Iran. One official summed up the Hegseth approach as: “Bombs away!”

Hegseth’s early exchanges with the generals made clear that the civilian harm program had no future in his reimagined “Department of War.”

“No more politically correct and overbearing rules of engagement,” Hegseth would later tell a gathering of hundreds of generals and admirals.

But now, after U.S. missile strikes have killed hundreds of civilians in Iran and Yemen, episodes that drew global criticism and questions from Congress, the program could soon be partially revived.

The about-face would mark a rare reversal in a department that hasn’t tolerated dissent under Hegseth and President Donald Trump. Last year’s military memos detail a delicate pushback that may preserve at least the framework of the civilian security campaign.

Current and former defense officials described the evolution of the civilian security program on condition of anonymity for fear of retaliation from the Trump administration.

The Pentagon did not respond to requests for comment about the poll or the program’s revival.

Casualties, Outrage

The first major military operation of Hegseth’s tenure — an air campaign in Yemen in the spring of 2025 — killed 153 civilians and wounded 243 others. The majority of the deaths came from just two strikes, ranking them among the deadliest for civilians the Pentagon has acknowledged in modern warfare, said Airwars, a British watchdog that tracks civilian harm from military operations. The toll was confirmed in the Pentagon’s 2025 annual report for civilian casualties, The Washington Post first reported.

Hegseth described the Yemen operation at the time as a “clear, limited mission executed with ruthlessness, full stop.”

By the time the U.S. began attacking Iran the next year, the Pentagon’s civilian protection program was down from a little under 200 people to around two dozen, some working only part time, the current and former defense officials said. Central Command, which said it needed a minimum of nine of its 16 staffers to function, was left with one person.

On the first day of the U.S. and Israel’s war on Iran, a U.S. airstrike hit a school in Minab, a city near the Strait of Hormuz, killing more than 100 children and igniting international outrage.

A colorful classroom with decorations for young children is in ruins after a bomb attack, with debris covering the floor.
The Shajarat al-Tayyiba Primary School is in ruins in Minab, Iran, on March 31. Hamid Vakili/Anadolu via Getty Images

Hegseth and the commander in charge of the Iran operation, Adm. Brad Cooper, were questioned about the strike before Congress. Hegseth said the U.S. does not target civilians and would investigate. Early reports on that investigation, yet to be publicly released, have said outdated U.S. intelligence led to the attack on the school.

Democratic senators and humanitarian groups questioned whether the cuts to the civilian protection personnel also contributed to the school tragedy.

In a letter to Hegseth in April, 11 senators said the school strike, as well as other deadly attacks on civilians in Iran, reflected “the administration’s broader disregard for the strategic, legal, and moral imperative to minimize civilian harm,” noting the dismantling of civilian harm reduction efforts. “This is a concerning pattern and raises questions about whether the administration is upholding international law and the laws of war,” it said.

Previously, a Pentagon spokesperson declined to answer questions from ProPublica about the letter, saying it would “respond directly” to Congress.

Concerns were voiced by military leaders before Hegseth largely eliminated the operation. One exception is the Department of the Army, headed by a civilian appointee, which agreed with getting rid of CHMR entirely.

Leaders at Central Command told Hegseth’s top staff in a memo that CHMR “enhanced our precision and accuracy, making us more lethal.” The memo added, “The operational and strategic value of this work is widely recognized.”

At the same time, the Joint Staff echoed the commanders’ sentiment. According to correspondence reviewed by ProPublica, the Joint Staff disagreed with the Army acting secretary, who said in a February 2025 memo that the program was redundant and recommended Congress repeal the law that set it in motion.

As pressure has built from both Congress and within the military to restore the program, the Defense Department appears to have backpedaled.

In April, Elbridge Colby, the Defense Department’s under secretary of policy, sent a follow-up memo about the civilian protection enterprise to Hegseth. This time, there was only one proposal on the table: reboot the program, albeit at reduced staff levels with 70 full-time people.

In that memo, Colby wrote that maintaining 35% of CHMR would address investigations into civilian harm, reports to Congress and payments to those harmed.

The same commanders Hegseth polled earlier endorsed the move, several defense officials told ProPublica.

The proposal would allow all regional commands to reconstitute their civilian harm reduction staff, and the Virginia-based Civilian Protection Center of Excellence for training, data and research would move from under the Army to the Joint Staff. The move, officials say, gives the mission more heft and eases collaboration with the Navy and Air Force, which are at the forefront of the administration’s campaigns.

“If not for the combatant commanders,” one defense official said, the program would be “dead.”

The harm mitigation effort makes up a tiny fraction of the Defense Department’s billions in spending. The center near the Pentagon had an annual budget of $7 million, officials said, with the cross-command total for all civilian security work estimated at around $80 million before the cuts. Officials expect a refashioned program to operate on a smaller budget.

Military leaders steeped in lessons learned from the post-9/11 “forever wars” repeatedly have affirmed harm mitigation as an asset on the battlefield. Civilian casualties undermine U.S. objectives and help boost insurgent recruiting, analysts say.

The Pentagon had formalized CHMR in 2022, detailing steps to protect noncombatants and investigate reports of casualties. Commanders had publicly championed the program in congressional testimony and defense journals. The current head of European Command, Gen. Alexus Grynkewich, co-wrote a paper calling it a “watershed” development.

In some cases the staff acted as a “red team” pushing back on strikes and the possible collateral damage. The center of excellence focused on institutionalizing best practices, training service members and collecting data on civilian harm.

Those efforts and “operational effectiveness are the same thing,” one retired general told ProPublica. “They’re two sides of the same coin.”

The post Top Commanders Told Hegseth Not to Cut Civilian Protection Program, Poll Showed. He Did It Anyway. appeared first on ProPublica.

❌