1. Keyra identity
Identity and trust powered by Keyra. Authentication is not authority.
2. Credential verification
Privileged roles require verified credentials. License numbers are not displayed without need.
3. Zero trust
Every protected request evaluates live authorization. UI hiding is never sufficient.
4. Encryption
TLS in transit. Encryption at rest in the deployed environment. Envelope encryption is prepared for highly sensitive fields. This page describes architecture; specific certifications are not claimed until validated.
5. Consent
Purpose-bound, time-bounded, revocable where operationally possible, and receipted.
6. Data minimization
Collect the minimum necessary. Do not put PHI in analytics, logs, error messages, URLs or page titles.
7. Audit logging
Append-only, hash-chained events. Historical audit records are never mutated.
8. AI provenance
Every material model output shows model, version, time, principal, task, sources and human review state.
9. Clinical governance
The platform never prescribes. Qualified clinical authority governs medical action.
10. Data sovereignty
Jurisdiction-aware residency. One region does not satisfy every law.
11. Model training policy
No automatic training on patient data. Independent, versioned choices.
12. Incident response
Safety operations queues exist for patient safety, unauthorized access, credential fraud and data exposure. Public incident SLAs will be published as operations mature.
13. Vulnerability reporting
Report suspected vulnerabilities through the contact channel. Do not attach patient records to reports.
14. Subprocessors
Infrastructure subprocessors will be listed here as they are contracted. None are implied by this software alone.
15. Policy version history
Every policy has a version, effective date and previous version.