North Carolina's public programs are adopting data analytics and AI. GDAC Watch tracks the buildout.
The 2026 Appropriations Act funds analytics, AI-assisted eligibility screening, fraud detection, and case support through the Government Data Analytics Center (GDAC, G.S. 143B-1385) and related state technology hubs — across SNAP, Medicaid, tax administration, child welfare, and statewide identity. Individually, these are program-integrity measures; together, they mark a significant expansion of data-driven decision support in public programs — one where governance practices are still maturing. GDAC Watch maps the buildout so the organizations that operate within it can see what's funded, what's live, what reports are due, and what to prepare for.
How the pieces fit together
Where each funded capability sits in a resident's interaction with state programs — from identity, to eligibility, to payment integrity.
Tracked provisions
The 2026 Appropriations Act footprint. Status pills update as systems move from funded to operational.
Watch calendar
Statutory dates, report deadlines, and effective dates. The next upcoming item is highlighted.
Event log
What changed, when, with confidence ratings. Newest first.
Case ledger: referenced litigation
Precedent and active cases that define what accountable automated decision-making looks like in public benefits and health coverage. Status as of September 26, 2026. Docket status changes; verify against the court record before relying on any entry.
| Case | What it concerns | Why it matters here | Status | Confidence |
|---|---|---|---|---|
| Public-benefits algorithms (the closest precedent for GDAC-era eligibility tools) | ||||
| K.W. ex rel. D.W. v. Armstrong, 789 F.3d 962 (9th Cir. 2015) | Idaho Medicaid budget-calculation tool for adults with developmental disabilities | Budget notices that did not explain why a person’s budget dropped were inadequate under due process. Automated outputs still owe people an explanation they can contest. | Decided (June 5, 2015) | HIGH |
| Ark. Dep’t of Human Servs. v. Ledgerwood, 2017 Ark. 308 | Arkansas ARChoices / RUGs assessment algorithm that cut attendant-care hours (reported at 43%) | Arkansas Supreme Court affirmed a TRO: likely violation of APA notice-and-comment when the algorithm-based rule was adopted. Process for adopting an automated rule is itself reviewable. | Decided (Nov. 9, 2017); later appeal dismissed as moot after DHS re-promulgated the rule (2019) | HIGH |
| Bauserman v. Unemployment Ins. Agency, 509 Mich. 673 (2022) | Michigan’s MiDAS system issuing automated fraud determinations without adequate process | Michigan Supreme Court recognized a damages remedy for due-process violations by an automated state system; class settlement later approved (Court of Claims, January 2024). | Decided; settled | MED |
| Health-coverage algorithms (utilization management) | ||||
| Estate of Lokken v. UnitedHealth Group, No. 0:23-cv-03514 (D. Minn.) | Alleged use of the nH Predict model to cut Medicare Advantage post-acute care | The leading test of whether an insurer’s reliance on a prediction model over clinical judgment is actionable. | Motion to dismiss granted in part; in discovery. Plaintiffs’ expert disclosures due October 14, 2026 | MED |
| Barrows v. Humana Inc., No. 3:23-cv-00654 (W.D. Ky.) | Same model, Humana Medicare Advantage post-acute denials | A parallel track on the same technology in a different circuit. | Proceeding after an August 15, 2025 ruling on the motion to dismiss; class certification motion due October 15, 2026 | MED |
| Kisting-Leung v. Cigna Corp., No. 2:23-cv-01477 (E.D. Cal.) | Cigna’s PxDx batch-review algorithm for claim denials | Tests whether batch algorithmic review satisfies the duty of individualized claim review. | Motion to dismiss granted in part and denied in part (March 31, 2025); active | MED |
| Adjacent health-AI accountability (consent, licensure) | ||||
| Saucedo v. Sharp HealthCare (Cal. Super. Ct., putative class) | Ambient AI scribe recording of clinical visits allegedly without consent | Consent and records-integrity claims under state medical-privacy and recording law, reportedly including chart entries stating consent was given. | Pending | MED |
| Lisota v. Heartland Dental (N.D. Ill., putative class) | AI recording and processing of patient communications allegedly without consent | Per-encounter statutory exposure for AI tools deployed without a consent process. | Pending (as reported) | MED |
| Pennsylvania v. Character Technologies, Inc. (Character.AI) | State action alleging a chatbot held itself out as a licensed medical professional | States are using existing licensing law to police AI, without waiting for AI-specific statutes. | Pending (as reported) | MED |
Sources: published opinions (Ninth Circuit; Arkansas Supreme Court; Michigan Supreme Court); Georgetown Health Care Litigation Tracker docket summaries (updated September 7 and 21, 2026); Baker Donelson, “AI Governance in Health Care” (June 8, 2026); contemporaneous reporting. MED means corroborated by secondary sources, pending review of the court record. No case in this ledger arises under North Carolina law; none yet exists on GDAC-funded tools. This ledger is educational, not legal advice.
The dashboard tracks what happened. The Quarterly Brief explains what it means.
The GDAC Watch Quarterly Brief is a confidence-scored analysis for organizations that operate within or alongside these systems — health plans, provider and supplier associations, legal and policy teams. Each issue includes:
- Independently audited analysis of the quarter's developments, with HIGH/MED/LOW/FLAG confidence mapping
- Implications by stakeholder: plans, providers, suppliers, counties, and the people these programs serve
- The forward calendar — what's due, what's effective, and what to prepare for before it lands
Opens an email to 5Q Health. Organizational subscriptions and briefing sessions available; independent analysis, not funded by any tracked entity.
Methodology & independence
Every item is verified against primary sources — bill text (including visual confirmation of amendatory strikethrough), session laws, agency reports, and contract records — using the 5Q three-pass discipline: draft, adversarial audit, final with confidence map. Ratings:
How to read this. GDAC Watch is a map, not a scorecard. Where it notes that safeguards are "not specified," that describes the statutory text as enacted — many of these programs are early-stage, consistent with how AI adoption is unfolding across states, and governance often matures through agency rulemaking and implementation. 5Q tracks that process to help every stakeholder — agencies, plans, providers, and the people these programs serve — navigate it well.
GDAC Watch is independently produced by 5Q Health LLC and accepts no funding from GDAC, its vendors, or any tracked agency. Corrections are logged, not deleted. Statutory citations reference the 2026 Appropriations Act as enacted (Session Law 2026-41); section numbers and dollar figures were re-verified against the chaptered session law on September 26, 2026. This tracker is educational analysis, not legal advice.