Two Pennsylvania men pleaded guilty after traveling repeatedly to Minneapolis to enroll vulnerable individuals in a Housing Stabilization Services program and billing roughly $3.5 million for services not provided. When investigators sought supporting documentation, the defendants used AI tools including ChatGPT to fabricate client notes and emails, according to the DOJ.

The Department of Justice announced on February 10, 2026, that two Pennsylvania residents admitted guilt in a scheme in which they traveled to Minneapolis to sign up vulnerable people for a Housing Stabilization Services program and then billed Medicaid for services that were never delivered, generating approximately $3.5 million in fraudulent claims. Investigators determined the defendants repeatedly recruited and enrolled clients, submitted false claims to the program, and, when challenged by auditors, produced fabricated documentation created with artificial intelligence tools such as ChatGPT. DOJ officials highlighted the case as notable because it involved explicit use of generative AI to manufacture clinical notes, emails, and other records intended to conceal the fraud. Multiple federal agencies assisted in the investigation, underscoring interagency cooperation on schemes that exploit both vulnerable populations and emerging technologies. The guilty pleas reflect growing enforcement attention on how AI can be misused to facilitate and conceal fraud against government benefit programs, and prosecutors indicated they will pursue appropriate criminal penalties and asset recovery.