Identity Security - AI Powered IAM and ITDR

This course is an "Identity Bootcamp", providing a progression from foundational identity best practices to advanced AI-driven implementation. It bridges the gap between proactive Identity and Access Management (IAM) and reactive Identity Threat Detection and Response (ITDR), using AI as the intelligence layer for real-time detection, behavioral analytics (UEBA), and automated response. The included LABs are built on a stack that supports Identity Orchestration, Behavioral Analytics, and Machine Identity Management.

The content is organized into 4 areas: The Human Element & Risk Orchestration, The Machine & Agentic Explosion, Identity Threat Detection & Response, and Privacy, Governance, and the Future. Zero Trust is the foundational philosophy of this course, woven into the curriculum by shifting focus from traditional perimeter security to "Identity-First" security.

Identity Security - AI Powered IAM and ITDR Benefits

• Course Benefits

  • Modernize authentication with FIDO2, WebAuthn, and passkeys, replacing vulnerable shared-secret methods.
  • Build AI-driven risk engines using behavioral biometrics for continuous identity verification.
  • Secure machine identities with SPIFFE and HashiCorp Vault using just-in-time credentials.
  • Govern agentic AI through security guardrails that control permissions and prevent data leakage.
  • Map identity attack paths and shadow administrators using graph analytics and BloodHound.
  • Automate threat response with Sigma rules and Wazuh-driven active defense playbooks.
  • Implement next-generation privacy using Zero-Knowledge Proofs (ZKP) and Decentralized Identifiers (DID).

  Prerequisites

  Attendees should have intermediate knowledge in networking and cybersecurity and knowledge of AI at the level of AI and Cyber Security: Attack and Defend.

AI Identity Security Training Outline

Learning Objectives

Module 1: Architecture of Modern Authentication

• Understand the shift from shared secrets to cryptographic identity verification.
• Analyze the limitations of SMS and TOTP-based MFA against modern attacks.
• Implement strong authentication using FIDO2 and WebAuthn.
• Deploy passkeys and passwordless authentication workflows.
• Secure account recovery with AI-assisted identity verification.
• Configure trusted certificate authorities and identity trust chains.
• LAB: Establish ADFS trust relationships using Microsoft PKI.
• LAB: Implement passwordless X.509 certificate-based authentication.

Module 2: Real-Time Risk Engines

• Building the AI "Brain" that decides when to trust a login signal
• Behavioral Biometrics: Capturing keystroke dynamics, mouse velocity and touch pressure
• Contextual Telemetry: Analyzing "Geo-velocity" and IP reputation signals
• ML Anomaly Detection: Training Scikit-learn models on "Normal" user login patterns
• LAB: Adaptive Risk Orchestration
• LAB: "Geo-Velocity" Risk Trigger
• LAB: Adding, executing and reviewing tests with Playwright

Module 3: Securing the Machine Workforce (NHI)

• Managing identities for the 95% of accounts that aren't human
• Workload Identity Federation: Understanding the SPIFFE standard for platform-agnostic identity
• Dynamic Secret Injection: Using HashiCorp Vault for "Just-in-Time" database credentials
• Attestation Mechanics: How containers prove integrity before receiving OAuth tokens
• Mutual TLS (mTLS): Securing the connection between Keycloak and autonomous AI agents
• API Security: Ensuring that autonomous agents have the correct OAuth "scopes" and "claims" before they can access backend data
• Auto-Enrollment AI Audit: Using AI to monitor AD CS logs for certificate anomalies
• LAB: Securing n8n workflows with mTLS
• LAB: mTLS with SPIRE Workload Attestation

Module 4: Identity Governance for Agentic AI

• Designing security guardrails for autonomous AI agents
• The "On-Behalf-Of" Problem: Manage how an AI agent proves it has explicit user consent to perform a task with OAuth
• Blast Radius Management: Restricting agent access based on intent with OAuth scopes
• Identity-Aware LLM Chains: OAuth tokens carry the identity context for AI prompts to prevent data leakage
• LAB: AI Agent Secret Retrieval
• LAB: AI Agent "Human-in-the-Loop" Approval

Module 5: Visualizing the Identity Attack Surface

• Using Graph Theory AI to see what the attacker sees
• Identity Graph Fundamentals: Mapping nodes (users) and edges (permissions)
• Shadow Admins: Detecting users with excessive permissions outside standard groups
• Tiered Administration: Implementing the "Red Forest" or Enterprise Access Model
• ESC (Escalation) Vulnerabilities: Using BloodHound to find Certificate Template misconfigurations
• Monitoring CA Logs: Sending AD CS "Certificate Issued" events to Wazuh
• LAB: The "BloodHound" Attack Path Hunt
• LAB: Detecting Certificate Forgery

Module 6: Active Defense & Autonomous Response

• Detecting and killing sessions in the middle of an attack
• Embody Zero Trust continuous monitoring and automated remediation
• Token Theft & Session Hijacking: Real-time detection of "Cookie Replay" attacks with OAuth bearer tokens
• Sigma for Identity: Writing rules for Kerberoasting, DCSync and Brute Force
• Automated Remediation Playbooks: Configuring Wazuh to trigger a "Global Logout" via OAuth APIs when an attack is detected
• LAB: Detecting "Golden Ticket" Anomalies

Module 7: Active Defense & Autonomous Response

• Using AI to automate the tedious parts of compliance
• Entitlement Outlier Detection: Using Peer Group Analysis to find excessive permissions
• Continuous Access Certification: Moving from quarterly reviews to real-time reviews
• Joiner-Mover-Leaver (JML) Pipeline: Automating role changes during department switches and termination
• CRL and OCSP: Using Keycloak to check if a Windows certificate has been revoked
• LAB: Entitlement Auditing with "Baton"
• LAB: The Offboarding Workflow

Module 8: The Future of Privacy: ZKP & DID

• Decoupling "Who you are" from "What you are allowed to do"
• Decentralized Identifiers (DID): Giving users ownership of their own identity "Wallet"
• Zero-Knowledge Proofs (ZKP): Mathematically proving a claim without revealing raw data
• Verifiable Credentials: Issuing digitally signed "badges" for offline verification
• LAB: Building a Zero Knowledge Proof Circuit