[CIF-K2F]
The Algorithmic Safety Net: Maryland’s Bet on AI in Human Services
Tier 2 — Systemic · 10 MAY 2026 · COGNOSCERE LLC
Maryland is deploying AI across SNAP, Medicaid, and child welfare faster than its governance rules. CIF Tier 2 systemic analysis of risks and implications.
Abstract
This Tier 2 Systemic brief, produced under the Cognoscere Intelligence Framework (CIF v7.8), examines the Maryland Department of Human Services’ adoption of artificial intelligence tools across its core human services programs, including SNAP, Medicaid, TANF, child welfare, and housing assistance. The analysis applies systemic-tier methodology to assess not only the deployment facts but the structural conditions that govern — and fail to govern — automated decision-making in one of the most legally consequential public-sector environments in the United States.
The primary finding is that Maryland has transitioned from exploratory to coordinated AI deployment faster than its governance architecture has matured. The state’s oversight mechanisms — including the Governor’s AI Subcabinet, the HB956 legislative workgroup, and the Department of Information Technology’s 2025 AI Enablement Strategy — are advisory in character and have produced no binding substantive restrictions on automated decision-making in high-stakes eligibility or child welfare functions. A parallel $1.2 million federally funded initiative, managed by Nava PBC, is developing open-source work-requirement verification tools intended to scale nationally through APHSA’s state agency network, effectively exporting Maryland’s governance assumptions before they have been tested by litigation or independent audit.
The significance of this case extends beyond Maryland’s borders. The combination of executive-level commitment, open-source tooling designed for replication, and an enterprise-wide AI infrastructure agreement covering 43,000 state employees positions Maryland as a de facto national template for AI in human services administration. The governance decisions the state makes — or fails to make — before the HB956 workgroup’s July 2026 report will shape the structural conditions under which millions of Americans interact with automated public benefits systems.
Researchers Also Ask
- What AI tools is Maryland using for SNAP and Medicaid eligibility decisions?
- How do states govern AI in public benefits administration?
- What are the due process risks of automated work-requirement verification?
- Is Maryland’s child welfare system using AI for risk scoring or removal decisions?
- How will federal work requirements under H.R. 1 affect state AI deployment in human services?
COGNOSCERE LLC · Structured Intelligence. Verified Sources. Decisions Supported.™