The U.S. Department of Housing and Urban Development is set to roll out an artificial-intelligence application designed to examine how its grant recipients allocate federal funds. Acting chief financial officer Irving Dennis, who has held the role since early 2025, said the system is intended to curb waste, fraud and abuse. HUD distributes roughly $77 billion in grants each year, yet officials acknowledge they currently lack detailed insight into post-award spending patterns.
A limited pilot involving ten grantees was described by Dennis as a “great success,” and HUD officials have slated a full launch for September 30. The agency is informing program offices of the upcoming policy shift and has hired a technical-assistance provider to deliver onboarding, training and troubleshooting. An internal review found no existing entry for the tool in HUD’s publicly reported AI inventories, though a source said the list is being updated.
Housing advocacy groups and policy think tanks voiced reservations about the initiative. Critics argue that an algorithmic review could introduce uncertainty into grant delivery, penalize smaller awardees lacking resources to contest AI-generated findings, and open the process to politically motivated decisions. FedScoop contacted dozens of national and local housing organizations, yet most were unaware of the pending tool and declined to comment when approached.
The AI capability will be embedded within the HUD Unified Grants System, known as HUGS, which recently secured a $500,000 contract with Palantir. Dennis said the department collaborated with a software firm and the Office of the Chief Information Officer to build the backend, while also obtaining funding from congressional appropriators and securing endorsement from the White House and senior agency leaders.
HUD officials emphasized that human oversight will remain central to grant-management decisions. A centralized data-science team inside the CFO’s office will monitor the AI’s output, continuously test models and follow a governance framework for analytics-driven choices. Darrell West, a senior fellow at Brookings’ Center for Technology Innovation, noted that this “human-in-the-loop” approach mirrors practices commonly adopted in the private sector.
Because HUD’s grants span state and local governments, nonprofits, for-profit firms, public housing authorities and tribal entities, the Office of Management and Budget classifies the AI tool as high-impact, triggering required risk-management controls. Officials cited examples such as one recipient spending a dollar per foot of lumber for disaster-recovery housing versus another paying ten dollars per foot for the same material, suggesting the system could flag cost inefficiencies.