AI Threat and Vulnerability Management
This topic covers AI threat modeling, misuse cases, adversarial inputs, data poisoning, model extraction, prompt abuse, vulnerability tracking, and remediation.
How to study for ISACA AAISM
Treat each item as a management decision: identify the AI asset and stakeholder, assess risk, select governance or control action, then document evidence and accountability.
Core concepts
Concept 1
AI threat modeling identifies misuse cases, attack paths, assets, trust boundaries, actors, impacts, and controls.
Exam cue: Use threat modeling before deployment or major design changes.
Concept 2
AI vulnerabilities may involve data, models, prompts, tools, APIs, integrations, identity, monitoring, and third-party components.
Exam cue: Treat prompts, data pipelines, tools, and model endpoints as attack surfaces.
Concept 3
Vulnerability management should prioritize remediation by exploitability, impact, exposure, business criticality, and compensating controls.
Exam cue: Prioritize remediation based on business impact and exposure.
Risk pitfalls and guardrails
Applying only traditional infrastructure scans to AI-specific attack surfaces.
Guardrail: Avoid treating AI security as only a technical scan, accepting risk without authority, or leaving prompts and data outside governance.
Ignoring data poisoning and prompt abuse in threat models.
Guardrail: Avoid treating AI security as only a technical scan, accepting risk without authority, or leaving prompts and data outside governance.
Prioritizing vulnerabilities only by scanner severity without AI context.
Guardrail: Avoid treating AI security as only a technical scan, accepting risk without authority, or leaving prompts and data outside governance.
Memory anchors
Threat Model
A threat model describes assets, actors, attack paths, trust boundaries, impacts, and controls.
Misuse Case
A misuse case describes how an AI system could be abused or made to fail.
Prompt Abuse
Prompt abuse attempts to manipulate instructions, context, tools, or outputs against policy.
Data Poisoning
Data poisoning manipulates training, tuning, or retrieval data to influence model behavior.
Model Extraction
Model extraction attempts to infer model behavior, data, or parameters through repeated interaction.
Attack Surface
Attack surface includes AI data, prompts, models, tools, APIs, identities, endpoints, and integrations.
Exploitability
Exploitability estimates how practical it is for a threat actor to use a weakness.
Remediation Priority
Remediation priority combines impact, exposure, exploitability, criticality, and available controls.
Checkpoint rule
Do the check-up only after you can summarize each concept in one sentence and identify one dangerous pitfall from memory.
Knowledge Check (after reading)
Short check-up to confirm understanding of this module.
Check-up Questions
An external feed can add training records without source authentication or anomaly checks. Which control is MOST appropriate?
Which evidence BEST demonstrates that controls over data poisoning operated throughout the review period?
Answer all questions to submit.
Next step personalized recommendations
Continue learning
Move forward only after this module is stable.
What is Pass Harbor?
Completely free exam prep for 317 U.S. exams.
- Practice questions
- Flashcards
- Study guides
- Mock exams
- No registration
- No paywall
- Start instantly
“No more expensive exam prep. Quality study tools should be accessible to everyone.”