In 2024, the U.S. Department of Homeland Security, or the DHS, accelerated its deployment of artificial intelligence technologies in the due process of immigration and border enforcement encompassed under the idea of broader national security. Through a series of new technology pilots, biometrics, and automated decision-making infrastructure software, the DHS claims to be modernizing its operations with the boom of AI for government operations, streamlining immigration vetting, and improving threat detection amid national security strategy. However, this rapid and uncalled expansion had triggered growing outrage among organizations for civil liberties and critics, as data privacy experts caution that such systems, when implemented at such a quick rate without a robust fundamental framework, can pose significant threats to due process, personal privacy, and constitutional rights.
Overview of the 2024 DHS AI Expansion
Guided by the 2024 DHS AI Strategy and aligned alongside the Biden Administration’s AI Executive Order and OMB’s M-24-10 directive on AI governance, DHS launched more than 50 advanced AI initiatives spanning Customs and Border Protection, Immigration and Customs Enforcement, the Transportation Security Administration, and U.S. Citizenship and Immigration Services which include facial recognition and computer vision tools located near transport hubs, biometric matching through multiple forms of individual data collection, predictive analytics for migration patterns and detection of any forms of suspicious activity, and natural language processing with sentiment analysis of asylum seekers and public media to name a few.
Privacy Risks: Scale, Surveillance, and Intrusion
The massive data infrastructure, which is necessary for the power of AI surveillance tools, makes the border region a site of unprecedented levels of data extraction. DHS collects a wide range of sensitive data and personally identifiable information, including travelers' biometric information, real-time GPS location, video feeds, online tracked activities, and perceived behavioral actions and cues.
Critics emphasize that such technologies for border security operate in a gray zone, which means that legal protections around search and seizure under the Fourth Amendment are weaker. In this 100-mile-long border zone where millions of Americans reside, analytics driven by watchlists, mobile phone data extraction, and automated license plate readers are used with limited oversight from judiciary bodies, which is a clear result of how the normalization of mass surveillance and neglect from the government impacts both migrants and U.S. citizens as well.
Compounding such concerns is the lacking of clear expiration timelines for stored data and removal of unwanted information for any specific purpose, insufficient safeguards on secondary data usage by third parties, and an opaque structure for partnerships with external vendors such as third party apps - many of whom use proprietary AI models which are difficult to audit or even contest in trials.
Algorithmic Bias and Discrimination
Facial recognition software and other such biometric algorithms have repeatedly been shown to produce higher rates of error for people of color, particularly Black, Latino, and indigenous individuals. In a border enforcement context, where split-second identification decisions can have profound effects on the individual’s life, as lies by its results in denial of entry, detention, or surveillance escalation, the danger of such biased outcomes is clear. Groups like the Electronic Frontier Foundation and the Center for Democracy and Technology have already flagged that AI decision-making, when based on previous enforcement data that had bias, can easily facilitate the reinstatement of discriminatory patterns. For instance, when such predictive algorithms are trained on the historical records of arrest or asylum refusal cases, which contain historical bias for certain sects and races, they can propagate systemic bias, further criminalizing certain communities or identities of color.
This can have a chilling effect on lawful migration policies and increase racial segregation while undermining the fair implementation of laws for immigration in the country. The lack of reliable mechanisms to appeal, explain, or correct algorithmic misinterpretations poses additional uncalculated risks for people, such as who are asylum seekers and non-citizens who are considered detainees and do not have access to adequate legal representation or protection.
Civil Society Response and Legal Challenges
In 2024, more than 140 civil rights organizations such as the ACLU, Just Futures Law and others collectively called for a moratorium of the use DHS had of AI in immigration enforcement as they had cited decades of racial surveillance and subjugation with the unprecedentedly high rates of error stemmed from emerging technologies and a dearth of clear avenues for accountability for the public media. Multiple lawsuits have followed back, and these have demanded the long-awaited details on how AI models are being developed, what datasets they are using, and how automated classifications impact immigration decisions as well.
Some lawmakers have also introduced legislation frameworks for regulating federal AI use for surveillance or immigration adjudication, as can be furthermore pinpointed and seen in acts such as the Protecting Individual Rights Against Government Use of Surveillance Technologies though such efforts have yet to have actually been passed through both of the chambers of the U.S. Congressional sections.
Innovation Balance with Rights
The increased use of AI across DHS represents a major change in the ways the federal government enables monitoring, management, and decision-making of mobility. Although technology can contribute to public safety and efficient immigration, unregulated use of it eventually will undermine and erode the underlying democratic and constitutional principles. Without enforceable protections and standards for transparency or independent oversight, such as AI systems will likely worsen discrimination, degrade privacy, and ability to instantiate injustice at scale.