# Joint Commission and CHAI on AI: what health systems must do

> Joint Commission and CHAI guidance (17 September 2025) and the voluntary 2026 certification: what they cover, what they do not certify, and CHAI's assurance labs.

Regulation · Regions: US · By Precision research team (Precision) · Published 2026-11-15 · Reviewed 2026-11-15 · Source: https://precision.example/learn/joint-commission-chai-responsible-use-of-ai

**TL;DR:** The Joint Commission and the Coalition for Health AI (CHAI) released voluntary guidance on responsible AI use on 17 September 2025. A voluntary organisation-level certification was reported launched in June 2026. It certifies how an organisation governs AI, not individual AI products, and CHAI's assurance-lab network was scrapped.

## Key takeaways

- The Joint Commission and CHAI released initial guidance on 17 September 2025; it is voluntary and covers policies, local validation, monitoring and use.
- A Responsible Use of AI in Healthcare certification was reported launched on 18 June 2026. It is voluntary, for US organisations, and certifies organisations, not products.
- The reported standard areas are governance, data management, risk and bias reduction, validating safety and effectiveness, and transparency and training.
- CHAI's plan for assurance labs was scrapped. Its registry and model-card repository continue, and vendors choose whether to publish.

## What did the Joint Commission and CHAI release?

On 17 September 2025 the Joint Commission and CHAI released initial guidance on responsible AI use in healthcare. It makes high-level recommendations on policies, local validation, monitoring and use, and CHAI said playbooks would follow. Both documents are voluntary.

## What is the 2026 certification?

A trade report dated 18 June 2026 says the Joint Commission launched a voluntary Responsible Use of AI in Healthcare certification, open to US organisations. It is organisation-level. The reported standards cover five areas:

| Area | What it asks about |
| --- | --- |
| Governance | Who is accountable for AI use |
| Data management | How data is handled and protected |
| Risk and bias reduction | How bias and harms are found and reduced |
| Validating safety and effectiveness | Whether tools were tested and monitored locally |
| Transparency, education and training | What users are told and taught |

Fees vary by organisation and we found no amounts.

## What happened to CHAI's assurance labs?

CHAI planned a network of assurance labs to evaluate AI tools independently. Fierce Healthcare coverage reports the plan was scrapped and later replaced by assurance resource providers. CHAI still runs a registry and model-card repository, built with Avanade, and vendors choose whether to publish.

## What does this mean for evaluation evidence?

Because the certification covers governance and not products, health systems must produce their own validation and monitoring evidence. The independent third-party validation role that assurance labs were meant to fill is open. See [the HTI-5 final rule](/learn/onc-hti-5-final-rule-decision-support) for the federal change that makes buyer demand more important.

## What did we not verify?

Fee amounts, formal certification requirements, uptake, and the date of the Fierce Healthcare article. Check the Joint Commission's page directly.

## Frequently asked questions

### Does Joint Commission certification certify an AI product?

No. The reported certification assesses an organisation's governance, data management, bias reduction, validation and monitoring, and transparency practices. It does not certify that any individual AI product is safe or effective.

### What happened to CHAI's assurance labs?

The plan for a network of assurance labs was scrapped, according to Fierce Healthcare coverage. CHAI later shifted to what it calls assurance resource providers. Its registry and model-card repository continue, with publication optional for vendors.

### Is this guidance mandatory?

No. The guidance and the certification are voluntary. They matter because they set expectations that health-system governance committees and buyers now cite when they ask vendors for local validation, bias testing and monitoring evidence.

> Limits of this page: The Joint Commission's own page returned an access error and the launch report comes from a trade-press source (DH Insights). Certification fees, formal requirements and uptake were not found. Details of the CHAI assurance-lab outcome come from search extracts of Fierce Healthcare; the article was not read.

## Where Precision fits

The 'monitoring and validation' area is where an independent evidence ledger helps: measurements, confidence intervals, change history and a monitoring plan, assembled as drafts your organisation attests to. https://precision.example/solutions/evidence-pack

## Sources

- Joint Commission and Coalition for Health AI release initial guidance, CHAI, 2025-09-17: https://www.chai.org/news/joint-commission-and-coalition-for-health-ai-chai-release-initial-guidance
- Joint Commission launches Responsible Use of AI certification, DH Insights, 2026-06-18: https://dhinsights.org/news/joint-commission-launches-responsible-ai-use-certification
- Inside CHAI's failed assurance labs, Fierce Healthcare: https://www.fiercehealthcare.com/ai-and-machine-learning/inside-chais-failed-assurance-labs
- CHAI creates central repository of AI model cards, Fierce Healthcare: https://www.fiercehealthcare.com/ai-and-machine-learning/chai-creates-central-repository-ai-model-cards
