The border agent stared at the computer screen, a knot forming in his stomach. The faces blurred together—he’d only had four weeks of online training before being thrust into the field. Had the AI hiring system made a mistake, or was he simply unprepared for the reality of immigration enforcement?
Immigration and Customs Enforcement (ICE) found itself in a bind, according to a recent NBC News report. An unknown number of undertrained law enforcement officers were deployed because the agency’s artificial intelligence system wasn’t working as intended.
President Trump’s promise of an unprecedented immigration crackdown preceded the hiring snafu. He had mandated a hiring blitz at ICE with the goal of 10,000 new recruits by the end of 2025, allocating $75 billion (€69.8 billion) to the agency.
The agency deployed an AI tool to speed up the hiring process. The system was designed to categorize submitted resumes, separating applicants into two groups. Those with previous experience as law enforcement officers were enrolled in a four-week online “LEO (Law Enforcement Officer) program.” Those without were supposed to attend an eight-week in-person course at the Federal Law Enforcement Training Center in Georgia, covering immigration law and firearm handling. Training used to be 20 weeks, before it was shortened to “cut redundancy and incorporate technology advancements,” per the Washington Post.
AI Hiring: When “Officer” Doesn’t Mean What You Think
I once saw a self-checkout machine mistake a bag of onions for apples. The AI hiring tool at ICE made a similar, if far more consequential, error. The NBC News report reveals that the AI tool misidentified applicants *without* law enforcement experience, sending them to the shorter “LEO program” in error. If the word “officer” appeared on the resume—compliance officers, for example—the AI flagged them as experienced law enforcement.
Two unnamed law enforcement officials told NBC they were unsure how many officers were improperly trained. Reportedly, the majority of new applicants were incorrectly flagged before the mistake was caught late last year.
What Happens When ICE Agents Use Excessive Force?
There’s been significant backlash to ICE’s increasingly aggressive enforcement tactics. These tactics have drawn criticism for targeting immigrants, both with and without documentation, and even American citizens.
Scrutiny intensified after the fatal shooting of Renee Nicole Good in Minneapolis. The agent involved had been with ICE for 10 years, according to NBC, and was therefore not subject to the AI screening. The problem here wasn’t lack of training, but something perhaps more ingrained.
ICE and AI: More Than Just Hiring
I’m starting to see a pattern. The hiring surge at ICE is only one piece of the puzzle; technology has been incorporated into nearly every enforcement mechanism. The agency has a contract with Paragon, an Israeli spyware maker, whose technology has been used to surveil journalists and migrant rights activists abroad. ICE uses an AI system for mass surveillance on social media, and agents have access to apps that scan irises and attempt to identify immigration status through facial recognition. The Department of Homeland Security even has its own AI chatbot called DHSChat. It was revealed in November that an ICE agent had used ChatGPT to compile a use-of-force report riddled with inconsistencies.
What Data Does ICE Collect?
Think of data as crude oil. ICE is drilling everywhere. The agent based the report solely on limited information and had ChatGPT make up the rest. The implications are staggering; if AI is fabricating justifications for use-of-force, what recourse do citizens have?
The over-reliance on automated systems has made ICE a blunt instrument. When algorithms misfire, the consequences can be devastating.
Can You Refuse to Answer ICE?
ICE is a dark mirror reflecting our anxieties about technology, immigration, and the balance between security and liberty. The promise of efficiency is alluring, but the potential for error—and abuse—is terrifying. As the use of AI expands in law enforcement, how do we safeguard against bias and ensure accountability?
If AI is left unchecked, could it erode the very principles it’s meant to protect?