The Technology Question at the Border
Immigration enforcement has always been a practical challenge as much as a policy one. In recent years, federal agencies managing border security have turned increasingly to artificial intelligence and advanced surveillance systems to manage the scale of the task. The question now before policymakers, and the public, is not whether these tools will be used—they already are—but how they should be governed, and what principles ought to guide their deployment.
This is fundamentally a conservative question. It asks how we balance security needs with respect for constitutional limits on government power, how we maintain human judgment in consequential decisions, and how we ensure accountability when technology operates at the edge of our understanding. These are not new concerns, but their application to modern surveillance and algorithmic decision-making requires clarity.
What These Systems Actually Do
Artificial intelligence and machine learning systems are already part of border security operations. These tools analyze vast amounts of data—vehicle patterns, crossing times, biometric information, travel histories—to identify potential security risks or irregular crossings. Some systems assist in real-time identification; others help prioritize which cases warrant human investigator attention. Border Patrol and Immigration and Customs Enforcement have deployed facial recognition technology, license plate readers, and predictive analytics systems designed to allocate enforcement resources more efficiently.
The appeal is straightforward: human officials cannot personally review every crossing, every vehicle, every piece of intelligence. Technology promises to extend their reach, catch patterns invisible to individual judgment, and reduce the burden of routine screening so that human expertise can focus on genuinely suspicious cases.
The Conservative Case for Caution
Conservative thought, properly understood, is not reflexively opposed to new technology or efficient government. But institutional conservatism does insist that we move carefully when expanding government power, especially into domains involving surveillance, identification, and detention decisions.
Several concerns deserve serious attention. First is the question of accuracy and bias. Algorithmic systems are only as reliable as their training data and design. If facial recognition systems perform differently across racial groups—a documented phenomenon in some applications—then deploying them at scale in immigration enforcement compounds rather than solves enforcement problems. It introduces error and the appearance of arbitrary decision-making into a system that already faces credibility challenges.
Second is the question of human oversight. The temptation with any efficient system is to let it run on its own. But consequential decisions about detention, deportation, or further investigation ought to involve human judgment, not merely human rubber-stamping of algorithmic recommendations. If AI systems are used to screen cases, the people making final decisions must understand how and why those recommendations were made.
Third is the transparency problem. Much AI system design is proprietary—companies and agencies keep their methodologies confidential. That may make business sense, but it creates a sovereignty problem when opaque systems make decisions affecting liberty and immigration status. Congress and the courts can oversee human decision-making; they have a much harder time overseeing what they cannot see.
What Governance Might Look Like
The answer is not to reject these tools outright. They can assist human judgment and improve resource allocation. But their use should be governed by clear principles that Congress, not agencies alone, should establish.
Those principles might include: systems used in consequential decisions must be auditable and subject to regular testing for accuracy disparities; human officials must review and approve all cases where AI tools recommend detention or denial of entry; agencies must maintain records explaining the basis for each algorithmic recommendation; private companies contracting to provide these systems should be subject to transparency requirements where government use is involved; and periodic review should assess whether systems are performing their intended function without creating disparate impacts on protected classes.
This is not radical restraint. It simply applies to algorithmic systems the accountability measures we apply to human decision-makers. It acknowledges that efficiency is good, but not at the expense of legitimacy and fairness.
The Institutional Question
There is a broader institutional principle at stake. Immigration enforcement is properly an executive function, but not an executive monopoly. Congress sets the rules; the courts review disputes; civil society—including technology advocates, civil rights organizations, and the press—scrutinizes practice. AI systems that operate in secret, that make no transparent record, or that allow agencies to avoid public accountability undermine that balance.
The Trump and Biden administrations have both invested in border technology systems. That suggests this is not a partisan impulse but a genuine trend. That is all the more reason for Congress to write clear rules now, before these systems become so embedded that changing them becomes politically impossible.
Border security is a legitimate government function. Efficient technology is a useful tool. But the legitimacy of immigration enforcement depends on the public's confidence that decisions are made fairly, accurately, and subject to meaningful oversight. Thoughtful governance of AI at the border is not about preventing enforcement; it is about ensuring that enforcement remains accountable to the law and the people it serves.
