Harness Health · Legal
SECURITY POSTURE · LAST UPDATED APRIL 26, 2026

How we secure the stack.

A short read on the architecture, the controls, and what we'll show you on a security review.

Architecture

Controls

The AI piece

Sage (the chat widget) is powered by Anthropic Claude. We use Anthropic's enterprise privacy terms — zero retention by default, no training on customer conversations. The system prompt for each brand is in our public registry at /kb/registry.json.

What we'll show you on a review

Reporting a vulnerability

Found something? Email security@harnesshealth.ai. We respond within 48 hours and credit responsible disclosure on the acknowledgments page.


For HIPAA-specific posture, see the HIPAA page.