AI
DOGE Used AI to Shape Housing Policy, But Government Refuses to Disclose Details
Federal officials deployed artificial intelligence to inform policy changes at HUD, yet the government is blocking public records requests about how the tools were used.
WIRED — AI
The Trump administration's Department of Government Efficiency (DOGE) leveraged AI systems to guide policy decisions at the Department of Housing and Urban Development, but federal officials are shielding records of that work from public view. According to documents revealed through a Freedom of Information Act request by the nonprofit legal group Democracy Forward, HUD is denying access to materials describing how AI tools informed agency decisions.
The DOGE Team at HUD#
Christopher Sweet, a recent economics graduate from the University of Chicago who worked at HUD, and Scott Langmack, a veteran of property technology companies, formed the core of DOGE's presence at the housing agency. Their mandate centered on using machine learning to flag existing regulations for potential elimination as part of a broader government-wide deregulation push. Langmack has since moved to the Office of Management and Budget, where he now directs deregulation AI efforts.
HUD staff initially reported that the AI system identified regulations for rescission, with agency employees then providing feedback on those flagged policies. Some staff members, however, questioned the value of the initiative.
The Missing Records#
Democracy Forward's request sought over 100 documents about HUD's use of artificial intelligence in decisionmaking. The agency withheld nearly all of them, invoking justifications that privacy advocates and legal experts find troubling. Beyond citing legitimate exemptions—such as protecting personal information or attorney-client privileges—HUD also asserted that certain materials qualified as "deliberative AI input" and fell under presidential communications privilege.
Among the concealed files were documents with telling titles: "GPT defined Econ Analysis approach 11 10 25.docx" and "RegulatoryAnalysisPrompt.pdf", both belonging to Langmack. Others bore labels suggesting they documented how the DOGE team constructed prompts to perform regulatory analysis. Several documents withheld for being part of the deliberative process indicated they related to regulatory review of different HUD programs, though the extent of AI involvement remains unclear.
Experts Question the Secrecy#
Privacy and government accountability advocates warn that the lack of transparency around AI's role in policymaking poses serious risks. Tori Noble, a staff attorney at the Electronic Frontier Foundation, emphasized that AI systems are prone to generating false information, perpetuating bias, and producing errors—making it essential to examine the exact instructions given to them.
