AI Security Training: Defend Against AI-Powered Threats

Certifying course — verifiable certificate included.

Master AI security with this vendor-neutral US program. Learn NIST AI RMF, EO 14110 & HIPAA through real-world scenarios. Apply skills immediately — enroll today.

149 USD

Online course, instant access after payment. 12 modules. Level : Advanced. Language : English.

Course outline

  1. AI-Powered Attack Vectors & Taxonomy — Mapping the Adversarial ML Threat Landscape Across Sectors
    • Identify and classify the full spectrum of adversarial ML attack vectors — including prompt injection, model inversion, evasion, model poisoning, and supply chain contamination — using a structured th
    • Analyze deepfake-based social engineering patterns across IT, finance, healthcare, and critical infrastructure sectors and articulate the distinct business impact of each attack type.
    • Apply a business-impact classification framework to map identified AI threats to organizational risk levels, producing a shared vocabulary for cross-functional security communication.
  2. MITRE ATLAS & AI Threat Intelligence Frameworks — Operationalizing ATLAS Tactics, Techniques & Procedures for AI Attacks
    • Navigate the full structure of MITRE ATLAS — including its tactics, techniques, and procedures (TTPs) specific to adversarial ML attacks — and correctly map at least three documented real-world AI inc
    • Correlate MITRE ATLAS with MITRE ATT&CK to build hybrid threat coverage that addresses both the traditional cyber kill chain and AI/ML-specific attack vectors, producing an integrated threat profile f
    • Source, evaluate, and operationalize AI/ML-specific threat intelligence from authoritative feeds and communities, and apply structured intelligence workflows to enrich SOC detection capabilities again
  3. Regulatory & Compliance Landscape for AI Security (US) — Navigating NIST AI RMF, EO 14110, HIPAA, and FFIEC Requirements
    • Apply the four core functions of the NIST AI Risk Management Framework (Govern, Map, Measure, Manage) to an organization's ML pipeline and identify concrete governance actions within each function.
    • Interpret the key provisions of Executive Order 14110, HIPAA AI implications, and FFIEC guidance and explain how each mandate shapes AI security obligations for US organizations across IT, finance, an
    • Execute a structured compliance gap analysis on a representative ML pipeline, producing a prioritized findings report that includes documentation and audit trail requirements for regulatory defensibil
  4. AI Risk Assessment Workshop — Threat Modeling an ML Pipeline & Producing a Structured AI Risk Register
    • Perform an end-to-end threat modeling exercise on a realistic ML pipeline by identifying high-value model assets, enumerating attack surfaces, and mapping threats to business impact categories.
    • Prioritize identified AI risks using a structured likelihood-versus-damage matrix, distinguishing critical threats from acceptable residual risks across IT, finance, healthcare, and critical infrastru
    • Produce a complete, structured AI risk register — including threat descriptions, asset owners, risk scores, and remediation actions — that can be adapted directly to any organizational ML deployment.
  5. Designing a Layered AI Security Architecture — Data Integrity, Model Governance & Runtime Monitoring Aligned to NIST AI RMF
    • Design a defense-in-depth AI security architecture that addresses data integrity, model governance, and runtime monitoring as distinct but interconnected layers.
    • Map each architectural control — including data provenance, role-based model access, and inference anomaly detection — to the corresponding NIST AI RMF core functions (Govern, Map, Measure, Manage).
    • Produce a secure MLOps pipeline design that integrates validation gates, versioning controls, and real-time monitoring into a continuous deployment workflow compliant with US regulatory requirements.
  6. Incident Response for AI-Driven Attacks — Adapting the NIST IR Lifecycle to Compromised Models — With a Timed Tabletop Exercise
    • Adapt the four phases of the NIST IR lifecycle — Preparation, Detection & Analysis, Containment/Eradication/Recovery, and Post-Incident Activity — to the unique characteristics of AI-driven attacks, i
    • Apply structured containment strategies for production models under active attack, including model isolation, traffic rerouting, and rollback procedures, and distinguish between eradication steps such
    • Execute a timed tabletop exercise to produce a complete, prioritized AI incident response plan under realistic time and information constraints, demonstrating decision-making discipline and cross-func
  7. Red Teaming AI Systems — Methods & Execution — Adversarial Input Crafting, Model Extraction & Structured Engagement Design
    • Craft and categorize adversarial inputs — including perturbation attacks, boundary probing, and jailbreaking techniques — and explain how each exploits specific vulnerabilities in ML model behavior.
    • Design a structured AI red team engagement from scratch: define scope, set rules of engagement, select appropriate tooling, and assign roles to ensure safe, reproducible, and legally defensible testin
    • Produce a professional prioritized vulnerability report that maps discovered weaknesses to business impact, assigns ownership, and provides actionable remediation recommendations aligned with NIST AI
  8. Red Team Lab — Hands-On Attack Simulation — Executing Live Attacks on AI Models & Drafting Remediation Recommendations
    • Execute prompt injection and evasion attacks against a live demo AI model and document observed model behaviors using a standardized severity scoring rubric.
    • Perform a query-based model extraction attempt and interpret the resulting confidence data to assess intellectual property and privacy exposure risk.
    • Draft a professional red team deliverable — a prioritized vulnerability report with remediation recommendations, assigned ownership, and realistic implementation timelines — ready for presentation to
  9. Building Anomaly Detection Models for Cyber Defense — From Feature Engineering to Real-Time Deployment in Defensive Pipelines
    • Select and justify the appropriate ML algorithm for a behavioral anomaly detection use case, mapping algorithm characteristics to specific threat types found in logs, network flows, and API telemetry.
    • Engineer a defensible feature set from raw security data sources, apply training and validation procedures, and tune hyperparameters to achieve a false-positive rate acceptable for SOC operations.
    • Compare batch scoring and real-time streaming inference architectures and recommend the optimal deployment pattern for a given organizational security posture and latency requirement.
  10. Automated Threat Intelligence with AI — NLP-Driven Feed Ingestion, IOC Enrichment & SIEM/SOAR Integration via STIX/TAXII
    • Ingest and normalize multi-source threat feeds using NLP techniques — including entity extraction, deduplication, and schema harmonization — to produce a unified, query-ready indicator dataset.
    • Design and implement a confidence scoring and automated IOC triage workflow that prioritizes high-fidelity indicators for analyst review and routes lower-confidence signals to automated suppression or
    • Configure end-to-end STIX/TAXII feed integration with enterprise SIEM and SOAR platforms, enabling automated IOC ingestion, alert correlation, and playbook-driven response within an operational SOC en
  11. Integrating AI Defenses into SOC Workflows — Reducing MTTD, Human-in-the-Loop Design & Measuring SOC Performance Gains
    • Map AI detection outputs — including anomaly scores, behavioral alerts, and threat intelligence enrichments — to structured SOC triage and escalation processes so that every AI-generated signal reache
    • Apply alert correlation, noise suppression, and priority queuing techniques within an AI-augmented SOC to measurably reduce Mean Time to Detect (MTTD) AI-specific threats while maintaining analyst foc
    • Design a human-in-the-loop governance model that incorporates analyst override mechanisms, structured feedback loops, and pre/post-AI integration performance metrics to continuously improve SOC effect
  12. Capstone — AI Security Program Design & Presentation — Full Architecture Design, Peer Review & Maturity Roadmap Debrief
    • Design and present a full AI security architecture for an assigned organizational scenario, integrating detection tools, an incident response plan, and governance controls into a coherent, defense-in-
    • Apply a structured peer review rubric to evaluate teammates' AI security designs, identifying gaps in coverage, technical feasibility, and regulatory alignment while offering prioritized, actionable f
    • Construct a forward-looking AI security maturity roadmap that synthesizes lessons learned across all three training days, translates identified gaps into phased improvement initiatives, and defines me

Access the course

See all our courses