← Back to Resources SECURITY

Security, privacy & compliance

How permissions, audit logs, encryption, and data handling work—so IT, risk, and compliance can evaluate Unphepha with confidence.

SECURITY PROGRAM

Controls aligned to enterprise reviews.

Unphepha is designed for organizations that must answer hard questions about data paths, identity, and accountability before enabling AI.

DATA HANDLING

What happens to your content.

Clarity beats buzzwords. Here’s the operating model security teams should expect.

1. Ingest

Connectors pull authorized content and metadata into your tenant’s knowledge store.

2. Protect

Encryption, ACLs, and sensitivity labels constrain access at rest and at query time.

3. Use

Search, chat, and automations only retrieve content the requesting identity can access.

4. Retain / delete

Retention schedules and deletion propagation keep the knowledge layer aligned to policy.

REVIEW PACK

Materials commonly requested.

Ask us for the pack that matches your stage—evaluation, procurement, or annual reassessment.

Architecture overview

Trust boundaries, major services, and data flow diagrams.

Questionnaire support

Help completing SIG/CAIQ-style vendor assessments.

Penetration & SDLC

How we develop, test, and remediate securely.

Subprocessors

Transparency on infrastructure and optional model providers.

FAQ

Security FAQ.

Straight answers for common blockers.

Is customer data used to train public models?

No. Customer content is not used to train public foundation models.

Can we keep data in a specific region?

Region options are available on Enterprise—confirm during security review.

How do we report an issue?

Use your support channel or security contact provided in your customer agreement.

Want this tailored to your organization?

Book a personalized demo and we’ll map this capability to your sources, teams, and controls.

Book a demo