Model data flows, actors, assets, attack surfaces, trust zones, and risk scenarios for private AI workloads.
Learning outcomes
Curriculum
The chapters are individually addressable documentation pages. You can link directly to a concept from another program, architecture decision, or implementation guide.
Define the system and its trust boundaries before describing threats.
Use threat modelling to identify credible failure, misuse, compromise, and abuse scenarios before selecting controls.
Set the intended purpose, actors, assets, dependencies, lifecycle, and system boundary before selecting controls.
Map where trust changes and where data, artifacts, requests, identities, tools, and evidence cross a boundary.
Identify what matters, where it is exposed, who or what can act, and how harm could occur.
Inventory the data, models, prompts, artifacts, identities, tools, interfaces, infrastructure, and evidence that require protection.
Model external, insider, supplier, user, model, and automated-agent actions without assuming one universal attacker.
Build traceable scenarios across ingestion, retrieval, inference, tools, supply chains, operations, and evidence.
Rank credible scenarios and connect them to testable controls, evidence, and residual-risk decisions.
Rank scenarios using explicit likelihood, impact, exposure, detectability, control strength, and uncertainty criteria.
Translate sovereignty goals into testable statements of what must be allowed, prevented, approved, observed, retained, and recovered.
Design evidence alongside controls so important decisions, releases, configurations, and operating events can be verified.
Complete a structured threat model and verify its traceability and decision quality.
Produce an asset inventory, attack-surface map, prioritized scenarios, control plan, evidence plan, and residual-risk register.
Verify that threat scenarios are complete, traceable, prioritized, and connected to controls and evidence.
Develop and prioritize abuse and failure scenarios for a private AI workload.
Design exercise; executable implementation pending
This public curriculum teaches a structured AI threat-modelling method. It is not a penetration test, production security approval, or guarantee that every threat has been identified. Practical assurance requires system-specific review and testing.