Healthcare AI content guide

ChatGPT Health is an independent physician led education and governance site founded by Harvey Castro, MD, MBA. It is not affiliated with OpenAI or the ChatGPT Health consumer product. It does not provide medical advice, diagnosis, or treatment.

Purpose

Help clinicians, health systems, patients, policymakers, and technology teams distinguish useful healthcare AI from unsupported claims and govern systems across the full lifecycle.

Core principles

1. Evaluate intended use and patient consequence, not model brand.
2. Separate vendor evidence, peer reviewed research, regulatory status, and local validation.
3. Match human oversight to consequence and preserve authority to override.
4. Test accuracy, calibration, failure modes, subgroup performance, accessibility, privacy, security, and workflow burden.
5. Monitor outcomes, drift, incidents, complaints, changes, and retirement criteria.
6. Publish corrections and review dates; do not treat a profile or badge as proof of safety or compliance.

Evidence record

The benefits and harms record includes randomized, prospective, observational, and external validation evidence. Each case explains the result, limitations, patient impact, and governance lesson without declaring any foundation model inherently good or bad.

Governance lifecycle

Inventory, classify, evaluate evidence, validate locally, authorize, deploy safely, monitor, and change or retire. Maintain an evidence profile with intended and excluded uses, version, data provenance, endpoints, subgroup results, limitations, oversight, privacy/security controls, monitoring thresholds, incidents, owner, and review date.

Canonical resources

• Homepage
• Evidence: benefits and harms
• Governance operating system
• External frameworks
• AI literacy
• Resources
• About
• Corrections and contact
• Founder profile

Last reviewed: September 28, 2026.