Back to SecureTheCloud Labs

AI Red Team Scenario Design - Complete L2 Track

AI Red Team Scenario Design L2 Track

Complete intermediate learning path for safely designing AI red-team scenarios without executing live attacks, abusing real systems, accessing customer data, or mutating production environments.

StatusComplete Track
Modules9 of 9
DomainAI Security
RuntimeRead-only course

Complete Module Map

1. AI Red Team Scenario Design Overview Implemented LAB - production quality-gated. Introduce safe AI red-team thinking, scenario boundaries, authorization scope, test objectives, and evidence expectations. 2. Prompt Injection Scenario Design Implemented LAB - production quality-gated. Design safe prompt-injection scenarios for instruction hierarchy, untrusted input handling, retrieved content boundaries, and refusal behavior. 3. Tool-Abuse Scenario Design Implemented LAB - production quality-gated. Model AI tool-use abuse, excessive authority, unsafe tool selection, approval bypass attempts, and tool permission boundaries. 4. Retrieval Poisoning Scenario Design Implemented LAB - production quality-gated. Review how poisoned, stale, low-authority, or tenant-crossing retrieved content can influence AI behavior. 5. Data Exfiltration Scenario Design Implemented LAB - production quality-gated. Frame data exposure scenarios using synthetic examples only, without real secrets, customer records, or credential material. 6. Agent Loop and Cost Abuse Scenario Design Implemented LAB - production quality-gated. Reason about runaway loops, repeated tool calls, expensive retrieval paths, quota abuse, retries, and operational cost risk. 7. Human Approval Bypass Scenario Design Implemented LAB - production quality-gated. Test whether AI workflows preserve approval gates, escalation paths, policy decisions, and separation of recommendation from execution. 8. Evidence Capture for AI Red Team Findings Implemented LAB - production quality-gated. Prepare reviewer-safe evidence: objective, preconditions, expected controls, observed behavior, findings, uncertainty, and remediation guidance. 9. AI Red Team Scenario Design Capstone Implemented LAB - production quality-gated. Combine scenario design, evidence capture, control mapping, risk explanation, and executive-ready reporting.

Governance Boundary

This is a complete, static, educational track. It does not run live red-team tests or connect to production systems.

Track implemented = true
Track complete = true
LAB modules implemented = 9 of 9
Backend exposure = false
Public backend exposed = false
Live model abuse execution = false
Live exploit execution = false
Customer data access = false
Credential handling = false
Runtime mutation = false
Production enforcement claim = false