Can You Pass the Pepper, Please?

A Lunchroom And A Report

Pedro Cantú Ríos was eating lunch when the room changed.

He was in the Anchorage Correctional Complex in Alaska, held in immigration custody inside a jail building, when guards fired plastic spheres into a communal space. The spheres burst into orange chemical dust. Cantú Ríos, who had a lung condition, told the Washington Post that the dust left him gasping. He put a towel over his face. “I thought I was going to die,” he said.

The official record has a different purpose. It does not begin with his lungs. It begins with orders, refusals, staff response, ventilation, and lockdown.

In the Alaska incident, according to the Post’s account of the internal records, some detainees were loudly cursing at staff, demanding their property, and refusing orders to return to cells. The incident report did not say the men acted violently or appeared on the verge of violence. DHS later said no detainee was directly struck by pepper-ball rounds and that staff opened a door for decontamination and ventilation.

Those two accounts do not cancel each other out.

A lunchroom can feel one way to the people responsible for bringing it under control and another way to the people breathing the air. A report can be accurate in some details and incomplete in what those details mean. A detainee can be frightened. A guard can be responding to a room that is getting harder to manage. An agency can believe its standards are being followed. The public record can be too thin to answer the questions that follow.

The gap between those accounts is where this story begins.

The cloud gives the story its image. The report gives it its civic problem.

The Washington Post obtained and analyzed internal ICE records called Daily Detainee Assault Reports. The records covered 1,460 use-of-force incidents at 98 ICE detention facilities from Jan. 1, 2024, through Feb. 28, 2026. They described punches, kicks, takedowns, restraint holds, restraint chairs, Tasers, pepper spray, and pepper balls.

During the first year of President Donald Trump’s second term, the Post found, detention staff used force 37 percent more often than the prior year. The number of people subjected to force rose 54 percent, reaching 1,330 individuals. The detained population at those facilities rose 45 percent over the same period, based on Relevant Research estimates used by the Post.

The records do not prove that every use of force was unlawful, punitive, or unnecessary. They show something more specific and more useful: civil immigration detention now operates at a scale where requests for property, water, food, medical care, and basic conditions can become compliance events, answered with jail tools, and preserved in reports that may not say enough for the public to understand what happened.

Civil Custody Inside Jail Walls

Immigration detention sits in an uneasy legal and practical space.

It is not supposed to be criminal punishment. People in ICE custody may be awaiting removal, asylum proceedings, bond hearings, transfer, paperwork, or court dates. Some have criminal convictions. Some have pending charges. Some have neither. The Post cited ICE data showing that 71 percent of detainees had no criminal convictions. ICE also describes immigration detention as non-punitive.

The buildings often operate like jails.

Many immigration detention sites are jails, former prisons, county lockups, private detention centers, converted facilities, or contractor-run buildings that inherit the routines of correctional administration. Doors lock. Counts happen. Movement is controlled. Meals, showers, medication, attorney visits, and property access happen on schedule. When a group refuses an order, the room can quickly become a security problem.

That does not make detention staff villains. It makes the setting important.

Crowded custody is hard to manage. People can harm officers, other detainees, or themselves. A group refusal can become dangerous. A staff member who waits too long may create risk. A staff member who acts too fast may create another kind of risk. Any serious account has to leave room for force in a real emergency.

ICE detention standards do the same. The Post reported that ICE standards authorize force only as a last resort, after reasonable efforts to resolve a situation have failed, and say force must not be used as punishment. The same reporting quoted DHS saying ICE officers are trained to use the minimum force necessary to resolve dangerous situations and receive de-escalation training.

The question is not whether officers may ever intervene.

The question is what happens when civil custody relies on jail tools to solve ordinary detention problems.

The Numbers In The Force File

The Post’s records are useful because they are not one story from one room.

They are a file: many incidents, many facilities, many summaries, many uses of the same official language.

The file has limits. The Post said its analysis is likely an undercount. The records covered facilities where 86 percent of ICE detainees were held last year, but 140 facilities did not appear in the reports, most of them county jails that contract with ICE to hold smaller numbers of detainees. The Post said it was unclear whether those facilities used no force or failed to report it.

Even with those limits, the file has weight.

The Post reviewed hundreds of Daily Detainee Assault Report emails summarizing reported staff uses of force at 98 facilities. Its methodology says reporters extracted dates, numbers of detainees involved, facility names, law-enforcement agencies, and whether the report indicated pepper spray was deployed. Reporters also used Google Pinpoint’s AI to identify possible injuries, then verified injury descriptions manually.

The record shows scale, and scale changes the question.

A single incident can be argued over as one room: what was said, who refused, who escalated, who had authority, who saw what, who breathed what. A national file asks a broader question. It asks whether the system has a pattern in how ordinary detention conflicts become security incidents.

The Post found that at least 106 detainees had been injured in use-of-force incidents since the beginning of 2024, with reports describing seizures, dislocated shoulders, broken arms, head injuries, and eye injuries. The paper also said the true number was probably higher because injuries were sometimes omitted from the reports.

That does not turn every officer into an abuser or every detainee into a passive victim. It does not answer each case. It does something more basic. It shows that the report form is not a clerical afterthought. It is one of the only public-facing ways a closed custody system explains how force entered a room.

When power becomes physical, the paperwork matters.

The report form can become a shield when its blanks are easier to fill than its facts.

The Words That Change The Event

A report does not merely record an event. It gives the event a name.

That is why the words matter.

A person can be described as asking, demanding, protesting, refusing, resisting, threatening, obstructing, or attacking. Each word points the reader toward a different explanation. A request for medical care sounds different from a refusal to enter a cell. A hunger strike sounds different from a disturbance. A pepper-ball deployment sounds different from orange chemical dust spreading through a lunchroom.

None of those word choices has to be dishonest to matter. A person can both demand medical care and refuse a count order. A group can both protest conditions and create a control problem. Staff can both try verbal direction and later use chemical agents. The record’s job is to preserve those distinctions rather than flatten them.

The phrase that does the most work in the Post story comes from the Stewart Detention Center in Georgia. In April 2025, according to one ICE report described by the Post, 35 detainees refused to enter their cells for count, saying they had not been seen by medical. After guards moved non-protesting detainees to the recreation yard, staff pepper sprayed those who remained. The reason given in the report was to “gain compliance and control of the pod.”

That is how a complaint can change shape.

A medical complaint becomes a count refusal.

A count refusal becomes disorder.

Disorder becomes a compliance problem.

A compliance problem becomes spray.

Spray becomes a report.

The point is not that the report is false. The point is that the report may be true and leave the civic question underdescribed. A reviewer needs to know more than whether people complied. A reviewer needs to know what they were asking for, how long the condition had existed, what staff tried before force, who authorized the response, whether medical personnel were consulted, whether video exists, what injuries were reported later, and whether the underlying complaint was ever addressed.

Compliance is a necessary word in custody. It helps staff describe whether an order was followed and whether a room was under control.

But compliance can also crowd out other words.

Food.

Water.

Medication.

Property.

Air.

When Complaints Become Compliance Problems

The Alaska lunchroom was not an isolated kind of scene. The records show the same basic turn again and again: a problem about property, food, water, medical care, or conditions becomes a problem about compliance.

In Alaska, the issue was property. Detainees wanted belongings. Staff saw refusal and disorder. Pepper balls were fired. Chemical dust spread. DHS later said no detainee was directly hit by the rounds and that a door was opened for ventilation. Cantú Ríos described the event through his body, not the report’s categories.

At Stewart, the issue was medical care. A group said they had not been seen by medical and refused to enter cells for count. DHS told the Post that staff worked with the detainees to gain voluntary compliance and used pepper spray only after eight hours. CoreCivic, which operates Stewart, said force in that incident was used in response to detainees refusing verbal directives and was not punishment.

At Torrance County Detention Facility in New Mexico, the scale was larger. The Post reported that guards used pepper spray on 65 detainees. A Cameroonian detainee who said he witnessed the March 2025 incident from a neighboring unit told the Post the men were holding a hunger strike over food and water conditions. DHS said the detainees refused lawful commands, became aggressive toward facility personnel, and that the incident was not related to water access. CoreCivic said force at Stewart and Torrance was applied because groups actively refused verbal directives, not because of peaceful requests.

Those responses belong in the story because they sharpen the central question.

Everyone agrees that force happened. The disagreement is about what kind of event force answered.

Was it a safety problem?

A conditions complaint?

A group refusal?

A protest?

A failure of medical access?

A staffing problem?

A control event?

The answer may be more than one thing. That is why thin records leave too much unresolved. They make the room easier to process and harder to understand.

Private Custody, Public Power

Contractors sit inside this story because much of immigration detention is carried out through public-private arrangements.

The Post found that most facilities reporting use-of-force incidents last year were run by private contractors. Those companies are responsible for hiring security guards, training them on ICE standards, and making sure their actions follow those rules, according to interviews with former ICE officials cited by the Post.

That fact should not become a slogan.

Private operation does not automatically explain every bad outcome. Public operation would not remove the basic difficulty of confined people, crowded spaces, legal uncertainty, medical needs, language barriers, transfer pressure, and institutional security. Contractors are not a single motive. ICE is not a single mind. Detainees are not a single type of person.

The pressure is structural.

ICE needs beds, movement, order, and compliance with federal standards. A contractor needs staffing, safety, contract performance, and daily control inside the building. A guard needs a room to settle before it turns. A detainee may need property, medication, information, water, food, counsel, or release from uncertainty.

By the time all of that reaches the form, it can be compressed into one word: compliance.

That is the public-power problem. A private company operating under federal standards is not exercising merely private judgment. It is carrying out custody in the name of the United States. When force is used, the record cannot belong only to the building. It belongs to the public system that authorized the confinement, paid for the bed, set the standards, and received the report.

Capacity, custody, and paperwork become one machine when the locked room writes its own account.

The Closed Door

Detention centers are hard to see.

That is not a metaphor. ICE detention facilities are closed to the public. The Post noted that ICE rarely releases surveillance videos from inside them. Unlike street enforcement actions, which may be recorded by bystanders with phones, detention force often happens behind doors where the main camera belongs to the facility and the main narrative belongs to the form.

That makes the report unusually important.

A report can protect staff when force is necessary. It can show why lesser steps failed. It can document threats, injuries, medical checks, video review, and supervisory approval. It can separate a lawful safety response from bad practice. It can also protect detainees by preserving facts they cannot easily carry out of the building.

The Post’s findings suggest that the reporting practice is not carrying that burden well enough.

Many reports used vague language or omitted key information, according to the Post. Some used phrases such as officers “guided” detainees to the ground, a wording the DHS Office for Civil Rights and Civil Liberties had criticized in a prior case as misleading. The Post also found that after Trump was sworn into office, facilities shifted toward shorter, more formulaic narratives. In 2025, the average narrative contained 40 words, about one-third as long as the year before.

That is a small administrative fact with large consequences.

A short report is not automatically a bad report. Some incidents are simple. Some details should be protected. Detainees have privacy interests. Investigations can involve sensitive information. Staff can also be unfairly judged if fragments are released without context.

A report that gets shorter while force events rise should draw attention. The more closed the room, the more careful the record has to be. Sparse language may help a report move quickly through an agency. It may also make the public record less able to carry the facts that matter.

The room is locked.

The form is what leaves it.

That means the form has to say enough.

What The Record Should Have To Say

A serious answer begins with record discipline.

Every use-of-force report should make the sequence visible. What was the immediate threat? What was the underlying complaint? What lesser steps were attempted? How long did the incident last? Who authorized force? Was medical staff consulted before force, if there was time? What medical review occurred afterward? Was video preserved? Were injuries reported later? Who reviewed the incident outside the direct chain involved in the event?

These are not anti-officer questions.

They are custody questions.

A staff member who made a reasonable decision in a difficult room benefits from a record that shows the decision clearly. A detainee who was harmed by unnecessary force benefits from a record that cannot hide behind vague verbs. A contractor benefits from standards that are not satisfied by formula. ICE benefits from a file that can survive outside review. The public benefits from knowing that civil detention does not become invisible when the door closes.

The second answer is pattern review.

If the same facility repeatedly uses chemical agents during disputes over food, water, medical care, property, showers, crowding, or confinement conditions, the inquiry should not stop at the last order refused. It should also ask what produced the refusal. That means examining staffing, medical access, grievance systems, crowding, training, language access, transfer pressure, and contractor incentives.

This is not a demand that every complaint be accepted at face value.

It is a demand that complaints not disappear merely because they became operationally inconvenient.

Custody will always need authority. Civil custody needs something more: a record that can show when authority was used because danger required it and when danger may have been created by unmet needs, bad conditions, poor communication, or a system stretched beyond its own standards.

The Paper After The Dust

The title sounds like a lunchroom joke until the pepper is in the air.

That is the point.

The word “pepper” belongs to ordinary life before it belongs to chemical force. It belongs to tables, trays, meals, appetite, habit. In the Alaska lunchroom, the ordinary word took on an official meaning. Pepper became a munition. Lunch became a control event. Breath became evidence that had to find its way into someone else’s paperwork.

No one in this story needs to be reduced to a role.

The people held in detention are not symbols. The officers and guards are not props. The agency is not a single motive. The contractors are not a single explanation. The record is not a confession. It is a public document trying to carry a closed-room event.

That pressure is the civic story.

A civil system reveals itself in the moment when an ordinary request enters a locked room and comes back as a compliance problem. Maybe the response was justified. Maybe it was excessive. Maybe it was both understandable and preventable. The only way to know is through a record that can bear more than the minimum.

The dust cleared faster than the paper.

The lunchroom leaves the public with a hard question, not an easy villain: when a closed room writes its own account, how much truth does the form have to carry?