Elephant Scale
Back to Certification

Cybersecurity for AI

Three certification levels from fundamentals to enterprise-scale AI security. Master threat modeling, prompt injection defense, and autonomous agent security.

Choose Your Certification Level

Three levels of AI Security certification, from fundamentals to enterprise-scale autonomous agent security.

Foundation

AI Security Essentials

Understand how AI systems are attacked before you defend them

Prerequisites

  • Basic programming (Python or Java)
  • Understanding of software development
  • Familiarity with APIs
  • Basic cybersecurity concepts

Topics Covered

  • AI vs Traditional Security: Why everything you know changes
  • OWASP Top 10 for LLM Applications — hands-on lab
  • AI Attack Surfaces: training data, APIs, model endpoints
  • Threat modeling fundamentals for AI systems
  • Input validation and output filtering for GenAI apps
  • Zero-trust basics applied to AI workloads
  • Monitoring & alerting for ML model anomalies

Exam Details

Duration:2 hours
Questions:60
Passing Score:70%
What You'll Achieve:
  • Map the full attack surface of any AI application
  • Apply OWASP LLM Top 10 mitigations immediately
  • Build your first AI threat model from scratch
Intermediate

Prompt Hacking & GenAI Defense

Learn to attack like a hacker, defend like a pro

Prerequisites

  • Foundation level OR equivalent
  • Python + LLM API experience
  • API security basics
  • Understanding of AI/ML concepts

Prerequisite: Foundation level or equivalent hands-on AI experience

Topics Covered

  • Prompt Injection deep-dive: direct, indirect, cross-agent attacks
  • Jailbreaking techniques: DAN, role-play, multi-language injection
  • System prompt extraction — live demo and countermeasures
  • OWASP LLM Top 10 exploitation in real apps — hands-on lab
  • Designing multi-layer guardrails: pre, in, and post-processing
  • Securing RAG pipelines and vector databases
  • AWS Bedrock / Azure AI Content Safety guardrail implementation
  • HarmBench red-teaming framework in practice

Exam Details

Duration:2 hours
Questions:60
Passing Score:70%
What You'll Achieve:
  • Execute and defend against prompt injection in production
  • Design guardrails that stop real attacks without breaking UX
  • Secure RAG pipelines and LLM API endpoints end-to-end
Advanced

Agentic AI Security & Governance

Secure autonomous AI agents at enterprise scale

Prerequisites

  • Intermediate level OR 2+ years AI/security engineering
  • LLM + cloud infrastructure experience
  • Enterprise security architecture knowledge
  • Compliance frameworks understanding

Prerequisite: Intermediate level or 2+ years professional AI security experience

Topics Covered

  • Agentic AI threat landscape: scheming, goal misalignment, tool abuse
  • Zero-trust architecture for autonomous AI agents
  • MITRE ATLAS framework — mapping AI attacks to enterprise systems
  • STRIDE-AI threat modeling for multi-agent systems
  • Adversarial ML: model extraction, membership inference, poisoning
  • MLOps security: secure CI/CD, container hardening, secrets rotation
  • EU AI Act, NIST RMF, ISO 42001 — compliance in practice
  • Enterprise AI governance and responsible AI security frameworks

Exam Details

Duration:2 hours
Questions:60
Passing Score:70%
What You'll Achieve:
  • Architect zero-trust security for autonomous AI agents
  • Build enterprise threat models using MITRE ATLAS and STRIDE-AI
  • Implement governance and compliance frameworks that pass audits
Study Guide

Exam Preparation Guide

Everything you need to prepare and pass your AI Security certification. Follow our structured approach for success.

Study Resources

Official documentation and hands-on practice tools

Official Documentation

OWASP Top 10 for LLM Applications

Complete vulnerability guide with examples

NIST AI Risk Management Framework

Official government guidelines

MITRE ATLAS Framework

Adversarial threat landscape for AI

EU AI Act

Regulatory requirements

Practice Labs

Prompt Injection Playground

Interactive attack simulations

RAG Security Testing Lab

Vector database vulnerability testing

AI Threat Modeling Tool

Practice creating threat models

Guardrail Implementation Lab

Build security controls

Exam Format

Question types and scoring breakdown

40%

Multiple Choice

Single and multiple correct answers testing theoretical knowledge

35%

Scenario-Based

Real-world security situations and problem-solving

25%

Practical Analysis

Configuration analysis and threat modeling

Level-Specific Preparation

Focus areas and study time for each certification level

Foundation Level

Focus areas and preparation timeline

Key Concepts
AI attack surface identification
OWASP LLM Top 10
Basic threat modeling
Input validation
Zero-trust principles
Study Timeline
Total Preparation20-30 hours
Recommended Period2 weeks
Hands-on Practice10-15 hours
Review & Mock Exams5-10 hours

Intermediate Level

Focus areas and preparation timeline

Key Concepts
Advanced prompt injection
Multi-layer guardrails
RAG pipeline security
System prompt defense
Cloud AI security
Red-teaming frameworks
Study Timeline
Total Preparation30-40 hours
Recommended Period3 weeks
Hands-on Practice20-25 hours
Review & Mock Exams10-15 hours

Advanced Level

Focus areas and preparation timeline

Key Concepts
Autonomous agent threats
MITRE ATLAS & STRIDE-AI
Adversarial ML
MLOps security
Enterprise governance
Compliance frameworks
Study Timeline
Total Preparation40-50 hours
Recommended Period4 weeks
Hands-on Practice30-35 hours
Review & Mock Exams10-15 hours

Test Day Preparation

Before, during, and after your exam

Before the Exam

  • Test webcam and microphone
  • Ensure stable internet
  • Clear workspace
  • Have government ID ready
  • Install ProctorWell software

During the Exam

  • Read questions carefully
  • Manage time (2 min/question)
  • Flag difficult questions
  • Stay focused on screen
  • Trust your preparation

After the Exam

  • Results appear immediately
  • Download certificate
  • Share on LinkedIn
  • Add to resume/CV
  • Plan next level

Exam Information

Everything you need to know about our certification exams.

Proctored Exams

AI-monitored via webcam and screen monitoring through ProctorWell

Instant Results

Get your results immediately upon completion. No waiting period.

Lifetime Validity

Your certification never expires and remains accessible forever.

Important Notes

  • • All exams are proctored via webcam and screen monitoring
  • • You must have a stable internet connection and quiet environment
  • • Results are available immediately upon completion
  • • You can retake any exam after 7 days if you don't pass
  • • Each level builds upon the previous one - we recommend taking them in order

Ready to Get Certified?

Take the next step in your AI security career with our industry-recognized certification.