Everyone Graduates ICE Killing Academy

Two out of three recruits who once would have washed out are now put in the field

In Part 1, I showed that 66.5% of ICE's surge hires have no discernible prior law enforcement experience—contradicting the administration's claim that 85% are seasoned professionals. But even if ICE is hiring people with no police background, maybe the training academy catches the problems? Maybe the accelerated 6 days a week for 8-week ICE Academy crash boot camp course at the Federal Law Enforcement Training Center (FLETC) weeds out those who should not carry a badge and a gun?

What do you think?

Among ICE officers who left during their academy tenure, 74% quit. Only 26% were dismissed by the institution. This is not a good sign.

Normal law enforcement academies show the opposite pattern—the institution removes more people than quit. At ICE, the pattern is inverted—early exits are disproportionately voluntary quits rather than institutional removals.

What Academies Are Supposed to Do

Law enforcement academies exist to screen out people who should not become officers. Physical fitness failures. Academic washouts. Conduct problems. Background issues that surface during training. The academy is the last institutional checkpoint before someone gets a badge, a gun, and the authority to use lethal force.

The Bureau of Justice Statistics (BJS) tracks academy outcomes nationally. In 2022, state and local academies showed a 14.6% non-completion rate. Of those who did not complete training:

• 56% were involuntarily removed (failed standards, conduct, background)
• 35% voluntarily quit
• ~6% other/unknown

The ratio matters: 1.6 involuntary removals for every voluntary quit. The institution does the filtering. That is the whole point.

While some may take issue with using state and local data for comparison, no federal law enforcement academy publishes such completion data, so the BJS baseline—drawn from 747 academies and over 60,000 recruits—is the best available benchmark for what professional law enforcement training looks like nationally.

What ICE's Academy Actually Does

I analyzed separation records for ICE Enforcement and Removal Operations officers who left with academy-period tenure—less than about 10 weeks of federal service, covering the compressed 8-week surge training program.

The pattern of how people leave the academy is upside down.

ICE ERO Separations with Academy-Period Tenure (Feb–Nov 2025)

Among separations with academy-period tenure during the surge, the quit-to-removal ratio was nearly 3 voluntary quits to 1 dismissal.

Compare to the BJS baseline of 0.6:1 (more removals than quits). ICE's 3:1 ratio is inverted by roughly five times.

The academy is not filtering—recruits who recognize they do not belong filter themselves out. And those who want to stay, the academy pretty much waves them on through.

Some quits may include pressured resignations—recruits "encouraged" to leave rather than be fired. We cannot know for certain because separation codes in the data do not distinguish truly voluntary departures from "resign or be fired." If such pressure is common, the 26% institutional removal figure understates actual filtering.

However...

The Attrition Rate Collapsed

It gets worse. Not only is the ratio of departures inverted, but overall attrition dropped.

In 2017, the DHS Office of Inspector General reported that ICE ERO's academy attrition rate was 12.2%. That was well before the surge, before the 8-week compressed program, before the mandate to hire 10,000 officers by year's end. (The OIG used cohort tracking; our cross-sectional method is not identical, but the comparison is directionally valid.)

The current attrition rate is one-third of the historical baseline—and this may well be an undercount, as recruits hired in late 2025 have not yet left or washed out. ICE is not screening more carefully during a hiring surge—it is screening less. Far less.

When NBC News reported in October 2025 that "200+ recruits" had been dismissed for failing fitness and academic standards, DHS disputed the concerns. But the overall picture is clear: attrition collapsed. The 200 dismissals are a fraction of the thousands who should have been filtered out if historical standards applied.

DHS says 85% of surge hires are experienced law enforcement officers who already completed accredited training programs so, of course, more make it through without administrative dismissal. But due to our data's limitations, we can only identify civilians with no discernible prior experience who are in training. We can only go by Length of Service (LOS), so if an agent has <0.2 years LOS, we assume that is a trainee. Our data keeps us from looking at former military or federal employees because even in training, with their credited service, we could not see them in the data.

It is civilians who are leaving the ICE academy—and they are walking out, not being shown the door. The ratio of quits to removals is inverted roughly five times from professional benchmarks.

If the civilian portion of the current workforce were being held to historical standards, we would expect attrition at least comparable to the 2017 baseline. Instead, two out of three recruits who would have washed out are now retained. The data is telling us a recruit needs to be particularly onerous to be forced to leave. Otherwise: here's your badge and gun.

Why This Happens

The math is straightforward. ICE was told to hire 10,000 officers by December 31, 2025. Training was compressed from 20 weeks to 8 weeks. The Spanish language requirement was eliminated. A second training facility opened in Charleston. The schedule went from five days a week to six.

Under these conditions, washing out recruits creates a numbers problem. Time is short and pressure enormous. Every dismissal is someone who needs to be replaced to hit the target--with an 8-week delay. The institutional incentive flips from "filter out problems" to "process volume."

Some might argue that high quit rates are fine—if recruits recognize they do not belong and leave voluntarily, the system is working. But this defense fails in context. Training was shortened. Requirements were reduced. And total attrition dropped from 12.2% to 4.2%. If the academy were a rigorous crash-course boot camp high-pressure operation, more people would leave, not fewer. The pattern is unmistakable: an easier program, fewer departures, and those who do leave are mostly walking out on their own accord.

The academy is now more turnstile than filter. And it shows.

The Compound Problem

Part 1 established: ICE is not hiring experienced officers. Two-thirds of surge hires have no discernible law enforcement background.

Part 2 establishes: ICE is not screening those hires during training. The academy attrition rate collapsed to one-third of historical levels, and the composition of departures inverted—self-selection replaced institutional filtering.

So who watches these under-trained officers with no prior law enforcement experience once they hit the field?

That is Part 3—Killing Supervision

Data Sources

All data is publicly available from the Office of Personnel Management:

FedScope/data.opm.gov: https://data.opm.gov (replacement for FedScope, which sunset January 28, 2026)

Separations Data: Monthly files, February–November 2025

Filters Applied:

• Agency: HSBB (Immigration and Customs Enforcement)
• Occupation: 1801 (Enforcement and Removal Operations)
• Supervisory Status: 8 (non-supervisory agents only)
• Academy Period: Length of Service < 0.20 years (~73 calendar days, covering the 8-week training period)
• Surge period defined as February–November 2025, based on hiring volume patterns in OPM data. An independent blind analysis identified September 2025 as the sharpest inflection point; our use of February is more conservative, capturing a longer period that dilutes rather than exaggerates the observed pattern.

Key External Sources:

• BJS State and Local Law Enforcement Training Academies and Recruits, 2022 (NCJ 309348, Table 5) — 56% involuntary / 35% voluntary baseline
• DHS OIG-19-07 (November 2018), p. 5 — 12.2% ERO academy attrition (2017)
• NBC News (October 22, 2025) — "200+ recruits" dismissed
• Newsweek (August 15, 2025) — Spanish requirement eliminated

Methodological Note: "Academy period" is defined as Length of Service < 0.20 years for surge hires, reflecting the compressed 8-week (48-day) training program. This captures fresh civilian hires only; veterans with bought-back military time would show higher LOS and are not captured in this analysis. Additionally, recruits hired in late 2025 may not yet have had time to wash out, so the 4.2% rate may undercount eventual attrition. The figures represent a floor on academy-period attrition, not a ceiling.

The 12.2% OIG baseline used cohort tracking (following specific classes through completion); our analysis is cross-sectional (point-in-time separations). The comparison is approximate, but the direction and magnitude of the difference—attrition collapsing to one-third—is clear.

An independent blind analysis of this data, using stricter thresholds (LOS < 0.15, September–November only), found a 2.81:1 quit-to-removal ratio—consistent with our findings. Our use of February as the surge start and a 0.20 LOS threshold is more conservative, capturing a longer period that dilutes rather than exaggerates the observed pattern. The BJS comparison is directional, not equivalent—different populations, different eras—but the magnitude of the inversion (nearly 5×) indicates a real pattern, not measurement noise.

Next: Part 3—"Killing Supervision"