CompTIA SecAI+ CY0-001 — Lesson 2: Data, Model, Prompt, Pipeline, and Supply-Chain Security

Learning Objectives

• Explain the roles, prerequisites, dependencies, and decision points shaping data, model, prompt, pipeline, and supply-chain security.
• Complete the workflow and produce AI pipeline security assessment that another practitioner can review.
• Resolve “untrusted retrieved content overrides application instructions” through a controlled and evidence-led troubleshooting process.

Detailed Topics

• training data integrity
• poisoning
• prompt injection
• model theft
• adversarial inputs
• pipeline access
• third-party models
• artifact provenance
• Security, performance, resilience, rollback, documentation, and support considerations for data, model, prompt, pipeline, and supply-chain security

Instructor Demonstration

The instructor demonstrates the complete data, model, prompt, pipeline, and supply-chain security workflow, records each material checkpoint in AI pipeline security assessment, and contrasts the accepted state with a controlled condition where untrusted retrieved content overrides application instructions.

Hands-On Exercise

Implement or analyze Data, Model, Prompt, Pipeline, and Supply-Chain Security in a controlled environment. Establish a baseline, reproduce or examine “untrusted retrieved content overrides application instructions,” isolate the proven cause, apply a reversible correction, and retain the accepted outcome in AI pipeline security assessment.

1. Confirm scope, access, dependencies, expected behavior, and safety constraints.
2. Execute the planned workflow while preserving configuration and operational checkpoints.
3. Test an expected path, a boundary condition, and the identified failure or exception.
4. Retest after correction and document rollback, ownership, escalation, and preventive action.

Scenario Analysis

• Identify the requirement with the greatest influence on data, model, prompt, pipeline, and supply-chain security.
• Compare two valid implementation or recovery choices using impact and evidence quality.
• Explain how the AI pipeline security assessment should be retained and transferred to the operational owner.

Troubleshooting Guidance

Start with effective configuration, recent changes, identities, connectivity, health, logs, and timing. Use AI pipeline security assessment to locate the earliest divergence between intended and observed behavior. Treat “untrusted retrieved content overrides application instructions” as a hypothesis until the corrected path passes the original test and a relevant boundary test.

Validation Checklist

• The implemented state for data, model, prompt, pipeline, and supply-chain security is traceable to approved requirements and inputs.
• The AI pipeline security assessment contains baseline, test, exception, correction, and accepted-result evidence.
• Security, data handling, capacity, performance, resilience, downstream impact, and rollback were considered where relevant.
• Monitoring, escalation, recovery, documentation, and support ownership are clear.

Lesson Outcome

Learners can apply Data, Model, Prompt, Pipeline, and Supply-Chain Security, explain the decisions made, and defend the result with AI pipeline security assessment.