ICE Reports 50,925 Arrests in August, a Third Straight Monthly Record; Share Without Convictions Rose in July Data

The Number Nobody Disputes, and the One Nobody Can Yet See

Two federal agencies published two ICE statistics twelve days apart, and neither one contradicts the other. That's the strange part. On September 2, 2026, the Department of Homeland Security announced that ICE had arrested 50,925 people in August — a new monthly record, and the third in a row^[1][2]. That works out to about 1,642 arrests every single day^[1].

Nobody disputes that number. What's disputed is a second number, one DHS didn't release at all. Case-level ICE records, pulled out through a Freedom of Information Act request and posted by the Deportation Data Project on August 24, show that in July, 73.5% of the people ICE arrested had no criminal conviction on their record^[3][9]. A year earlier, that share was 62.6%^[3].

Put those two facts side by side and you get the whole fight. The agency is arresting more people than ever. And a rising share of them have never been convicted of a crime. Both things are true at once, and which one you lead with tells you almost everything about who's telling the story.

Three Buckets, One Word Doing All the Work

To understand why this argument won't resolve, you need to know how ICE actually sorts the people it arrests. The agency uses three categories: people with a criminal conviction, people with a pending criminal charge they haven't been convicted of, and people whose only violation is a civil immigration one — no crime, no charge, just an immigration status problem^[12][13].

That third category is doing enormous political work. Critics call people in it "noncriminal." The White House says that's the wrong word, because entering or staying in the country without authorization is itself against federal law — a misdemeanor the first time, a felony after that^[6]. So when someone's only violation is a civil immigration one, the administration's allies argue they've still broken a law, just not one that shows up as a criminal conviction^[6].

That distinction matters because "conviction" is a narrower, more specific claim than "criminal record." A pending charge is an accusation a court hasn't ruled on. A civil immigration violation isn't a crime at all under how ICE's own categories work. When the Washington Post reports rising numbers of people "with no criminal record," and the underlying data actually measure "no criminal conviction," those aren't quite the same claim — and the gap between them is exactly where the White House pushes back^[4][5].

The administration adds a further wrinkle: it says nearly 70% of the people ICE arrests have been convicted or charged with a crime in the United States, and that the real number of public-safety risks is even higher, because many arrestees have foreign criminal records that American databases simply don't capture^[5]. There's no way for reporters, courts, or Congress to check that claim independently, since the records in question — if they exist — aren't in any U.S. system^[5].

Why the Math Almost Guarantees This Outcome

Here's the part underneath the argument that neither side really disputes: if you scale up any enforcement operation, the mix of people you catch will change, almost mechanically. People with serious convictions are the hardest to find. Many are already known to law enforcement or already in custody somewhere else. People with no record at all are often the easiest to locate — at a job site, outside a courthouse, at a home address already on file^[3][7].

So when ICE's monthly arrest target goes up, the share of arrests made up of easier, lower-level cases tends to rise too, regardless of what the stated priority is. Restrictionist analysts, including the Center for Immigration Studies, argue this is simply what successful expansion looks like: the pool of already-convicted, already-locatable people was worked first, and it's finite^[6]. A falling conviction share isn't proof of mission drift, in this reading — it's the natural residue of scaling up.

Immigrant-rights groups and researchers read the same trend the opposite way. The Vera Institute and independent researcher Austin Kocher both point to the growth in ICE's overall detained population — which topped 65,000 people by mid-July 2026 — and argue the overwhelming share of that growth is coming from people with no conviction at all^[7][8]. To them, that's the promise of targeting "the worst of the worst" quietly giving way to a numbers-driven quota.

There's also a timing problem baked into how this data becomes public. DHS puts out its monthly arrest total within days, by press release, often paired with named examples of serious crimes^[1]. The breakdown of who's actually being arrested — the conviction data — comes out much later, through FOIA litigation and outside researchers, and right now only runs through early August^[9]. That means August's record total is fully documented. August's composition is still mostly a guess based on July.

What TRAC's Numbers Actually Show

One further data point helps ground the debate: TRAC, the nonpartisan Transactional Records Access Clearinghouse at Syracuse University, reports that as of July 11, 2026, 46,436 of the 65,765 people held in ICE detention — 70.6% — had no criminal conviction^[18]. That's a snapshot of who's in custody at one moment, not a monthly arrest tally, but it lines up with the same broader pattern: a large and growing share of the detained population has never been convicted of a crime.

Meanwhile, the removal side of the system hasn't kept pace with the arrest side. ICE reported more than 350,000 removals in fiscal year 2026 as of July^[17]. Arrests can be scaled up fast. Removals require immigration court capacity and, often, cooperation from another country — neither of which moves quickly. That gap is filling detention beds and stretching out how long people stay in custody^[16][17].

The agency doing all this arresting is also, at the moment, without a confirmed leader. On September 18, 2026, the White House withdrew Lance Schroyer's nomination to head ICE^[10]. DHS itself has had new leadership since Markwayne Mullin was confirmed as secretary in a 54-45 Senate vote on March 24, 2026, replacing Kristi Noem^[15].

How the Coverage Split

Outlets on the right largely led with the record total and DHS's own examples of serious crimes tied to specific arrests — DUI manslaughter, child sexual abuse, fentanyl distribution^[1][2][11]. Fox News ran a piece built around the White House's rebuttal to the "noncriminal" framing, presenting the 70% convicted-or-charged figure without flagging that it measures something broader than the conviction-only statistic it was answering^[5]. The Washington Times used more careful language, reporting that ICE "claims" the record — a small but real hedge that signals the figure is the agency's own^[11].

Outlets on the left, including the Washington Post and NBC News, led with composition rather than volume: more people with no criminal record arrested in 2026 than in all of 2025, and July marking the first month of this term when people with only civil immigration violations made up a majority of arrests^[3][4]. Their sourcing is the same FOIA'd case data everyone else eventually cites, but their framing occasionally uses "no criminal record" more loosely than the underlying "no conviction" measurement actually supports^[4].

Al Jazeera folded the arrest numbers into a broader accountability narrative, tying them to Schroyer's withdrawn nomination, and cited a slightly different July total — 49,571, versus DHS's own 50,208 — without explaining that the gap reflects two different record-keeping systems, agency press releases versus case-level data^[1][10].

What almost none of the coverage disputes is the arithmetic itself. What it disputes is which of ICE's three categories should count as "criminal," and who gets to make that call. That question is still open, and the next data release — whenever it clears the FOIA pipeline — will be the next chapter, not the last one.