ICE arrests set a record in July 2026, and for the first time most of the people arrested had no criminal record

The number, and where it came from

ICE arrested 49,571 people in July. NPR's Vanessa Romo reported the figure on August 26. It works out to about 1,599 arrests a day, and it beats June's record of 43,021 by roughly fifteen percent. The Guardian, reporting the same data the same day, called it the highest monthly total since October 2022. Removals rose alongside it, with nearly 34,000 people removed in July against a stated White House target of 2,000 arrests a day.

The data did not come from a press release. It came from ICE, released to the Deportation Data Project, a research effort at UC Berkeley Law and the UCLA Center for Immigration Law and Policy that obtains the agency’s records through Freedom of Information Act litigation. That matters for reading everything below. Nearly every figure that describes who is being arrested, rather than how many, exists because two law schools went to court for it.

The tactics changed before the numbers did

Homeland Security Secretary Markwayne Mullin took the job in March. NPR describes the shift plainly. He dropped his predecessor’s camera-facing raids and replaced them with volume. Graeme Blair, a co-director of the Deportation Data Project, told NPR where the arrests are now happening. “They’re arresting people at airports, at ICE check-ins, at immigration courthouses and this is the result of that.”

Melissa Shepard, who directs legal services at the California immigrant services group ImmDef, described the same change to the Guardian from the other side of it. “The way that people are being detained is far less flashy, but we are still continuing to see arrests,” she said. Mullin has been open about the method. He called the 287(g) program, which deputizes state and local officers for immigration work, “the greatest partnership we can have,” and described his goal as enforcement in a “more quiet way.” That program held 1,372 agreements in January. By late August ICE listed 2,382, across 39 states and two territories.

July was the month the majority had no record

This is the finding that matters most for the families we evaluate, and it is the one the government disputes. The Guardian’s analysis found that July was the first month of this administration in which a majority of people arrested, 51 percent, had no criminal convictions and no pending charges. Not no violent conviction. Nothing. The Washington Post traced the same curve on August 28: detained immigrants with no convictions and no pending charges rose from 3,165 in February 2025 to 25,193 in January 2026.

The local numbers run sharper. NBC Bay Area reported more than 330 arrests in Bay Area cities in July, a 76 percent jump over the same month last year, with about 189 of them in San Francisco. Roughly 83 percent of the San Francisco arrests involved people with no past convictions and no pending charges.

ICE disputes the data. The agency told NBC it cannot verify “the source of, accuracy, completeness, or methodology of the data,” and said the Deportation Data Project “is not accurate.” It then said nearly seventy percent of its arrests involve people “charged or convicted of a crime in the U.S.,” without providing supporting data or naming a time period. Two things are worth holding onto there. The records in question are ICE’s own, produced by ICE in response to a lawsuit. And the seventy percent figure turns on the word “charged.” A charge is not a conviction, and an immigration violation can itself be the charge. A number quoted without its definition has stopped describing anything.

What it looks like where families live

In Danbury, Connecticut, NBC Connecticut recorded what appeared to be an ICE team detaining a man near City Hall on August 27, and local advocates counted roughly fifty people detained over the week. Senator Chris Murphy said ICE was arresting parents at school bus stops. A DHS spokesperson answered that any such claim “is FALSE,” and said the arrest in question involved a man taken in a car, not near a bus stop. We are not in a position to resolve that dispute and will not pretend otherwise. The second-order effect is not in dispute. Kate Dias of the Connecticut Education Association described families weighing whether to walk to the bus stop at all, in the first week of school.

In the Bay Area the pattern is quieter, and for our purposes it is the more important one. Milli Atkinson, who runs the San Francisco Bar Association’s Immigrant Legal Defense Program, gave NBC the local shape of it in one line. “People going voluntarily to their hearings, to their appointments, to their interview, and being detained.” Many of the people picked up at airports, she noted, still hold lawful presence. A work permit. An application on file. Compliance used to be protective. In July it was a location.

Status is being removed faster than people are being arrested

The arrest curve is not just reaching people who never had status. After the Supreme Court cleared the way to end Temporary Protected Status designations, the Guardian reports protections pulled from 50,000 Haitians and 6,100 Syrians, a first deportation flight to Haiti since the decision, and enforcement concentrated around Springfield, Ohio, where advocacy groups say Haitians were called in to ICE offices and fitted with ankle monitors. Salvadoran TPS is set to expire in September, which would end protections for another 170,000 people. Shepard’s read on what comes next was one sentence. “I think we are at the precipice.”

Then, on August 26, the State Department paused immigrant visa interviews at embassies and consulates worldwide. The stated reason is officer training on public charge guidance, so that consular officers evaluate consistently whether an applicant is likely to become a public charge. Reuters and the Wall Street Journal both reported it. Pre-interview processing continues, nonimmigrant visas do not appear to be covered, and no resumption date has been announced. Set that beside the USCIS public charge guidance that takes effect September 18, which we covered in the August briefing, and the direction is the same on both sides of the consular window. Household finances are being read more closely.

What this changes about the evaluation

Anticipatory fear is no longer speculative, and that is a clinical distinction, not a rhetorical one. When a qualifying relative describes lying awake over whether a spouse will come home, that is a statement about a documented environment. It is not a catastrophic cognition to be gently corrected. Anxiety proportionate to a real and quantified risk is assessed differently than anxiety that has outrun its trigger, and it is documented differently in a report an adjudicator will read.

Reassurance based on compliance has stopped working. For years the reasonable thing to tell a frightened family was that people who attend their appointments and keep their paperwork current are not the ones who get taken. The Bay Area reporting says otherwise, and so does the airport pattern. Families now have direct evidence against that reassurance, so offering it in an interview costs credibility that the rest of the session needs.

Children are witnesses now, not just dependents. A child who watched a parent taken from a car has a trauma exposure with a date and a place, and it belongs in the record with the same specificity as anything else. Scheduling has also gotten harder in a particular way: detention can happen at a check-in, which means an evaluee can become unreachable between intake and interview with no warning. Tell us early when a client has an upcoming check-in, hearing, or flight, and we will sequence the interview around it.

For VAWA, U visa, and T visa petitioners the calculus deserves saying plainly. These are people who came forward. They are inside a legal process, often with work authorization, and the reporting shows that a pending case is not shelter. Some will ask whether filing was a mistake. That question surfaces in an evaluation session, and it should be documented as what it is, a rational safety appraisal rather than avoidance.

None of this moves the extreme hardship standard, the VAWA elements, or our protocol. The clinical interview, the standardized instruments, the collateral review, the report, and our clinical lead’s co-signature are unchanged. What the data changes is context, and context is evidence when it is sourced.

General information about immigration policy and the psychosocial evaluation process. Not legal advice, and not a substitute for consultation with an immigration attorney.