Related videos:
An error in the use of artificial intelligence (AI) within the United States Immigration and Customs Enforcement (ICE) resulted in numerous new recruits being sent to field offices without receiving proper training, as revealed by NBC News, citing two law enforcement officials familiar with the case.
The problem arose as ICE was ramping up an ambitious plan to hire 10,000 new agents, using an AI tool to sort through job applications and determine which candidates had prior experience in law enforcement.
An error in the classification of the recruits
The automated tool was designed to identify candidates with police experience and assign them to the LEO (Law Enforcement Officer) program, a condensed course that requires four weeks of online training.
However, the AI made a crucial mistake: it incorrectly classified individuals who were not law enforcement officers simply because the word "officer" appeared in their resumes.
According to one of the officials cited by NBC News, individuals were included in the short program who were described as "compliance officers" or even people who merely expressed their aspiration to become ICE agents, without ever having been part of local or federal law enforcement.
"They were using AI to scan the resumes and discovered that many of the individuals identified as LEO were not," explained one of the officials.
Insufficient training for sensitive tasks
Candidates without police experience should have completed an eight-week in-person training at the ICE Academy, located at the Federal Law Enforcement Training Center (FLETC) in Georgia. This program includes training in immigration laws, firearms handling, operational procedures, and physical fitness testing.
Due to the error, a significant number of recruits received only reduced training, despite not meeting the necessary requirements, and were sent to field offices.
Although officials clarified that local ICE offices usually provide additional training before deploying agents, they acknowledged that the failure posed a serious risk in a highly sensitive operational context.
Late corrections and lack of official response
The error was detected in mid-autumn, when more than a month had already passed since the start of the expedited hiring process. After identifying the issue, ICE began manually reviewing the resumes and implementing corrective measures.
“Now they have to bring them back to the FLETC,” one of the officials indicated, referring to the need to complete the full in-person training.
The Department of Homeland Security (DHS), under which ICE operates, did not respond to requests for comment from NBC News.
Concerns about the use of AI in security
The case once again highlights concerns about the use of artificial intelligence in critical processes, particularly in agencies responsible for policing and immigration functions. An excessive reliance on automated systems, without adequate human oversight, can lead to errors with operational and security implications.
As ICE continues to refine its hiring process, the episode highlights the risks of prioritizing speed over thoroughness in an agency that handles detentions, deportations, and the use of force as part of its daily functions.
Filed under: