Deception Engineering and High-Confidence Signals
Learn how decoys create controlled security claims, why a rare trigger still needs context, and how deception can produc...
3 lessons · 2 hr 30 minCourse
Detection Engineering for AI and Agentic Systems
About this learning content: Courses, lessons, assessments, explanations and illustrations may be created with the help of artificial intelligence. We review and check the material and do our best to avoid incorrect or outdated information, but mistakes, omissions or ambiguous questions may remain. Please verify information before relying on it for professional, security, legal or operational decisions. Read the full notice or report an issue.
Difficulty
Advanced
Modules
3
Lessons
3
Time
2 hr 30 min
Language
en
Created by
Threat Intelligence Lab
Updated
16 September 2026
About this course
Understand how identity, delegated authority, tool use, memory, and data movement create evidence in AI systems that can act on a user’s behalf. The course is written for one reader and develops its ideas through continuous explanation, realistic thought experiments, and explicit distinctions between fact, assessment, and uncertainty.
What you'll learn
- ✓ Reason from a security concern to evidence and a bounded analytical claim.
- ✓ Explain assumptions, tradeoffs, and limitations in plain language.
Before you begin
You will get more from this course if these foundations are already familiar.
- Detection Engineering: From Telemetry to Tuned Detections — You should understand the basic detection lifecycle and common telemetry sources.
Course content
Module 1: Understand the Agent as an Authority Chain
Follow the user, application, model, agent, credential, tool, and resource identities that make an action possible.
Module 2: Observe Tool Use Without Logging Everything
Design evidence for tool calls, model interactions, retrieval, memory, and policy decisions while respecting privacy and proportionality.
Module 3: Detect Boundary Crossing and Respond Safely
Recognize unauthorized goal expansion, dangerous tool sequences, and policy bypass while keeping automated response proportional to certainty.