The Trump Administration’s Shift on AI Transparency
The Trump administration has seemingly erased evidence of artificial intelligence tools from government databases, tools that could significantly affect individual rights, following an investigation by the Monitor. One such AI tool, labeled as “high-impact” by the Department of Health and Human Services (HHS), was in development to assess sponsors of unaccompanied immigrant children in government custody. Under congressional mandates, federal agencies are required to publish comprehensive information about their AI tools online.
Recently removed from HHS’s website, the product description categorized this AI tool as “computer vision.” This field of AI allows machines to analyze extensive digital imagery and video to draw conclusions. Technologies within this realm include facial recognition software, license plate readers, and systems that analyze surveillance footage from drones.
Transparency Issues Amid Congressional Requirements
Currently, it remains uncertain whether a deal has been finalized or if this AI tool has been deployed, as HHS has not responded to multiple inquiries seeking clarification. Following a report on July 10 listing the AI tool in HHS’s online inventory, the Monitor sought further information from the agency’s Office of Public Affairs but received no response. Remarkably, ten days later, the AI product’s information had vanished from the site.
According to congressional requirements, HHS must maintain a public registry detailing the government’s “current and planned uses” of AI, known as the AI Use Case Inventory. Legal experts, including former government officials, express concerns that removing information regarding AI tools may violate the law and raises serious questions about the administration’s commitment to the welfare of migrant children.
Concerns from Privacy Experts
“The actions of HHS suggest a possible violation of the law, creating an impression of concealment,” remarked Jeramie Scott, senior counsel at the Electronic Privacy Information Center. Deirdre Mulligan, faculty director at the Law and Technology Center at UC Berkeley, who previously directed the Office of the National Artificial Intelligence Initiative under the Biden administration, emphasizes that such actions warrant congressional oversight.
The Office of Management and Budget (OMB) issued guidance last year requiring agencies to catalog all AI use cases, whether they are pre-launched, piloted, deployed, or retired. OMB’s directive insists that AI products remain listed for at least a year, even if their development or use is ultimately terminated.
The Role of High-Impact AI Tools
AI tools classified as “high-impact”—those serving as the primary basis for decisions materially affecting rights or security—are subject to rigorous scrutiny. This includes independent assessments of their implications for privacy, civil rights, and civil liberties. Furthermore, high-impact applications necessitate a challenge process that incorporates human review.
Before its removal, information on the Unaccompanied Child Sponsor Identity AI tool was limited online. The instructions emphasized that ensuring a child’s welfare mandates that potential sponsors undergo thorough vetting to confirm their identities. Advocates for immigrant rights are concerned that data from AI-driven screenings might be shared with other agencies to fulfill deportation objectives, potentially displacing children from their most suitable caregivers.
Implications for Child Welfare and Immigration Policy
As part of its enforcement efforts, HHS collaborates with Immigration and Customs Enforcement (ICE) and other federal and local agencies to secure the well-being of children in its care. Fingerprinting and other procedures for sponsors have been mandated under the Trump administration, which has led to thousands of arrests. Reports indicate that unaccompanied minors are now retained for prolonged periods, averaging over six months, due to heightened scrutiny of their sponsors.
Samuel Bagenstos, who was HHS’s general counsel from 2022 to 2024, posits that regardless of immigration status, authorities should encourage sponsors who prioritize children’s welfare to come forward without fear. However, potential sponsors may refrain from doing so, as coming forward could trigger ICE investigations. Historically, HHS has not operated under a law enforcement framework.
Debate Surrounding AI Use in Immigration
Bagenstos stresses that government agencies must adhere to OMB directives, questioning whether HHS’s actions align with the law. Trump administration officials, along with Republican lawmakers, have frequently claimed rampant immigration fraud and accused the Biden administration of failing to safeguard unaccompanied children. Yet advocates argue that these concerns are exaggerated.
Advocates for unaccompanied minors express apprehension over the potential sharing of data with the Department of Homeland Security for enforcement purposes, prioritizing immigration enforcement over child safety. Losing a viable sponsor could result in a child remaining in government care for an extended duration. Under President Trump, the Department of Homeland Security has initiated at least 51 “computer vision” projects, with over a dozen employing facial recognition technology aimed at bolstering immigration enforcement.
