Thursday, September 3, 2026

1. Signal of the Day · SNF / Post-Acute

   Skilled nursing AI adoption nearly doubles, but a quarter of facilities report no savings at all

   McKnight's 2026 Mood of the Market survey of 578 nursing home owners, executives, administrators and nurse managers finds 27.9% now call AI already useful, up from 17.1% a year ago — yet 25.8% report no actual savings in any of twelve categories, rising to 32.4% among nurses. The split between administrative and clinical adoption is the finding: 32.6% of administrators call AI already useful against 19.5% of nurse leaders, and the AI actually deployed across post-acute today is admissions, HR, claims and revenue cycle rather than clinical tooling. SNF and post-acute operators building AI governance should note that a program scoped only to clinical AI would currently govern almost nothing.

   HighBy Kimberly Marselas · McKnight'sRead at source ↗

2. SNF / Post-Acute

   A practitioner proposal: pay nursing homes for AI adoption through Payroll-Based Journal

   A long-term care regulatory educator argues that because the FY2027 skilled nursing payment update applies a 0.9% productivity reduction against a 3.3% market basket increase — netting 2.4% — on the assumption that technology raises productivity, CMS should consider recognizing AI adoption through a Payroll-Based Journal job code tied to star ratings. The column is opinion rather than policy, but it articulates a real structural grievance: post-acute absorbs a productivity adjustment premised on technology while operating on the margins least able to fund it. It also usefully bounds the automation question, noting resident interviews and hands-on oral and skin assessments cannot currently be AI-completed.

   LowBy Joel VanEaton · McKnight'sRead at source ↗

3. Regulation & Policy

   Consumer groups urge CMS to drop expanded provider-expulsion authority from home health rule

   Consumer Voice, the Center for Medicare Advocacy and four other organizations filed an August 28 comment letter arguing CMS already holds sufficient authority to act against chronically poor-performing nursing home owners, and asking the agency to strip expanded disenrollment and payment-clawback provisions from the CY2027 Home Health rule in favor of a standalone rulemaking. The provision matters to post-acute operators, but it carries no artificial intelligence or algorithmic component — a distinction worth preserving, since it is easily conflated with the separate CMS payment-integrity work that does rely on machine learning models.

   LowBy Kimberly Marselas · McKnight'sRead at source ↗

4. Security & AI Risk

   AI agent security startup launches with $50M, says it filters out 27% of publicly available agent skills

   AIR emerged from stealth with $50 million across two seed rounds to discover AI agents running inside enterprises, continuously re-vet the skills, plug-ins and MCP servers those agents load, and block ones failing security criteria. The company says roughly 27% of the add-ons and skills it finds online are filtered out, and its backers describe the problem as continuous re-verification rather than scanning: a previously approved skill could become risky if a package it downloads changes or its developer's account is compromised, so approval is a continuous re-verification duty rather than a one-time gate. Healthcare technology leaders deploying Copilot or agent extensions should ask whether anyone maintains an inventory of installed agent extensions and re-checks them after initial approval. (Corrected September 15; see Standards & Corrections.)

   MedBy Ram Iyer · TechCrunchRead at source ↗

5. SNF / Post-Acute

   Advocates report staff-worn consumer AI glasses recording nursing home residents without consent

   California Advocates for Nursing Home Reform reports that residents at several California facilities have observed staff, including nurses, wearing Ray-Ban Meta AI glasses and declining to remove them on request, with video recorded in areas containing protected health information and without prior consent. A single voice command records an entire shift and feeds it to the glasses' AI companion; the only indicator is a small light that is easily covered. Employee-owned wearable AI sits outside the controls operators normally rely on — no business associate agreement, no procurement review, no device management — while implicating resident dignity rights, state recording-consent law and HIPAA simultaneously. Prevalence is unquantified and the state industry association says no tracking data exists.

   MedBy JW August · OB RagRead at source ↗

Get the Daily Dispatch by email · free, about 8 a.m. Eastern · RSS