Skip to content
Cloud SecurityDeskSearch
Menu

Publications / Research note

Research note · AI systems

Threat-model the system around the model

The model endpoint is one component. The consequential paths often run through retrieval stores, orchestration identities, evaluation data, and operator tools.

Key findings

  1. The orchestration identity often spans more systems than the model endpoint.
  2. Retrieval and evaluation data need separate provenance boundaries.
  3. Operator tools can turn model output into privileged action.

Demonstration publication. The scenario and all numerical data are illustrative, not observed research findings.

By
Umair Akbar and Ahmed Elshekh
Published
Reading time
6 minutes
Coverage
AWS · Azure · Google Cloud

A concise research note for extending cloud threat models around production AI workloads.

Follow the action path

Start with what the application can cause, not only what the model can produce. Map each tool call, queue, function, data store, and human approval that turns output into an effect.

Separate data roles

Training, retrieval, evaluation, conversation, and operational telemetry serve different purposes. Give each a named owner, provenance expectation, retention decision, and permitted set of consumers.

Questions for the review

Ask questions that connect model behavior to cloud control evidence.

  • Which identity performs retrieval?
  • Where are tool arguments logged?
  • Can an operator replay a sensitive request?
  • What stops output from invoking an unintended action?

References

  1. NIST AI Risk Management Framework
  2. OWASP Top 10 for LLM Applications

From the desk

About the authors

Every Cloud Security Desk publication is authored by Umair Akbar and Ahmed Elshekh, the publication’s owners and chief editors.

Owner & Chief Editor

Umair Akbar

Editorial biography forthcoming.

Owner & Chief Editor

Ahmed Elshekh

Editorial biography forthcoming.