AI Security Hub
03AI Security Pillar

Context and Retrieval Control

Control which documents, records, and knowledge sources an AI system can access and use.

Business focus

Limit what AI can retrieve

Why it matters

Security starts with a defined business boundary

Context and retrieval controls apply existing permissions to AI search, keep client information separated, and preserve evidence about the sources used in an answer.

Business risk

An AI assistant should not gain broader access than the person using it. Weak retrieval controls can expose another client's documents, stale procedures, restricted records, or deleted information.

What this pillar covers
  • Document filtering and approved sources
  • Role-based retrieval authorization
  • RAG source validation and provenance
  • Context isolation between users or clients
Operating model

Turn limit what ai can retrieve into repeatable controls

A policy is only the starting point. For each AI use case, name the business owner, define the allowed boundary, configure the relevant technical controls, and decide what evidence proves those controls are working. Repeat the review when the model, data, connected tools, or business purpose changes.

Start with a single high-value workflow instead of trying to govern every experimental use at once. That makes it possible to test the controls with real users, find exceptions, and create a pattern the rest of the business can reuse.

StepDecisionEvidence to retain
1Scope the workflowOwner, purpose, approved data, users, and connected systems.
2Apply the controlsConfiguration, access rules, approval points, and test cases.
3Operate and reviewLogs, review results, exceptions, incidents, and change records.
Launch reading list

Start with these practical guides

Each pillar begins with one anchor guide and two supporting articles. Published guides become active automatically as they enter the blog.

Available

RAG Security: How to Protect the Documents Behind Business AI

Treat retrieval as a governed data-access workflow rather than only an accuracy feature.

Read the guide
Available

Vector Database Security: Protecting the Data Behind AI Search

Protect embeddings, metadata, credentials, indexes, and retrieval permissions.

Read the guide
Available

AI Data Provenance and Lineage: Building a Traceable Evidence Trail

Record where AI data came from, how it changed, and where it was used.

Read the guide

Questions for leadership

  • Does retrieval enforce the user's existing permissions?
  • How are client and tenant records kept separate?
  • Can a reviewer identify the sources used for an answer?
  • How quickly do deletions and permission changes reach the AI index?
Practical control checklist

Put the pillar into practice

  1. 1Approve each source before adding it to retrieval.
  2. 2Enforce authorization at query time, not only during ingestion.
  3. 3Attach ownership, sensitivity, tenant, and retention metadata.
  4. 4Test cross-client and cross-role isolation.
  5. 5Log retrieved sources and remove stale or deleted content.
Authoritative guidance

Use recognised guidance to validate the control design

These resources help teams translate AI-specific risks into documented, testable business and technical controls. Apply them to the actual data, permissions, and actions in the workflow rather than treating them as a one-time compliance exercise.

Review a real workflow

Turn this pillar into operating controls

Map the data, access, approvals, monitoring, and evidence around one important AI use case before expanding it.

Explore AI services