The Joint Commission's AI certification, answered
The Responsible Use of AI in Healthcare certification was announced on 1 June 2026 and its standards manual is not yet public — but it is being awarded: Hackensack Meridian Health earned the first RUAIH certification on 29 July 2026. These pages answer what can be answered from the published record, say plainly where the record stops, and label inference as inference. That discipline is the whole point of reading them.
Updated 31 August 2026. New: the realistic RUAIH certification timeline, and our founder's essay on the evidence mechanics ran on KevinMD.
The questions
Does RUAIH cover ambient clinical documentation?
No published RUAIH text names ambient documentation, and the Joint Commission is explicit that the programme certifies an organisation's gov…
Does RUAIH require an AI governance committee?
The standards manual is not publicly available, so no requirement can be quoted — but every source behind the certification treats an accoun…
How to prepare for RUAIH — the order that matters
Order preparation by what cannot be rushed later. First the AI inventory (everything depends on it), then the governance committee (decision…
Preparing for a RUAIH survey — the record beats the binder
RUAIH survey mechanics are not yet published, but Joint Commission surveys rhyme across programmes: an assessor, a sample, and 'show me' req…
RUAIH certification requirements — what is known
The RUAIH standards manual is unpublished, so no one can quote a requirements checklist — anyone who does is guessing. What the record suppo…
RUAIH evidence — the artifacts an assessor asks to see
Fifteen artifacts cover the published consensus, from the AI use-case registry and committee charter through local-validation protocols, mon…
RUAIH vs CHAI — how the two actually relate
They are layers, not competitors. CHAI is a coalition that publishes consensus guidance (eight playbooks, May 2026); The Joint Commission is…
RUAIH vs NIST AI RMF — certification versus framework
The NIST AI RMF (1.0) is a voluntary reference framework — four functions (GOVERN, MAP, MEASURE, MANAGE) applicable to any AI in any sector.…
The first RUAIH certification — what it actually tells us
On 29 July 2026, Hackensack Meridian Health became the first system to earn RUAIH certification. Three things follow: it is real and earnabl…
The five RUAIH focus areas, explained
RUAIH's announcement names five standards areas: (1) Governance, (2) Effective data management, (3) Risk and bias reduction, (4) Monitoring,…
The RUAIH timeline — dates that are actually published
Four dates now anchor the timeline: September 2025 (the joint guidance), 28 May 2026 (CHAI's playbooks), 1 June 2026 (the RUAIH announcement…
What does RUAIH certification cost?
The Joint Commission has not published fees for RUAIH, so any specific number you read is invented. The plannable costs are preparation cost…
What is RUAIH certification?
RUAIH — Responsible Use of AI in Healthcare — is a voluntary certification programme announced by The Joint Commission on 1 June 2026, built…
Who needs RUAIH certification?
As announced, RUAIH is voluntary — no law or accreditation condition requires it. Most reason to pursue it early: systems deploying clinical…
The five focus areas
RUAIH focus area 1 — Governance
RUAIH focus area 2 — Effective data management
RUAIH focus area 3 — Risk and bias reduction
RUAIH focus area 4 — Monitoring and validation
The glossary
The terms the certification conversation uses, defined the way an operator needs them defined.
- AUROC — the number on the baseline
- Demographic parity gap — fairness as a number
- Drift attestation — the quarterly signature
- Human in the loop — oversight as design
- Model card — the label on the model
- Model drift — how models fail quietly
- Override rate — the oversight vital sign
- Performance baseline — the locked number
- Risk tiering — proportionate governance
- Sensitivity and specificity — the two error rates
- Shadow mode — running without acting
- Survey binder — one model's whole story