Overview
C|OASP is the first EC-Council credential built specifically for offensive AI security. It validates the ability to red-team AI systems, execute adversarial attacks against large language models and agentic AI architectures, and build defenses that hold against real-world threats. The most advanced and intensive course in the catalog, with a performance-based exam component requiring demonstrated skill in simulated environments.
Prerequisites
Recommended: 3 years of cybersecurity experience; suited for penetration testers, red team operators, SOC analysts, and AppSec engineers.
Objectives
- Offensive AI methodology: MITRE ATLAS alignment, AI attack taxonomy, OWASP LLM/ML Top 10 (2025)
- AI reconnaissance and attack surface mapping: OSINT for AI assets, model/API/vector store enumeration
- AI vulnerability scanning and fuzzing: tools and techniques for models, pipelines, and deployments
- Prompt injection and LLM application attacks: jailbreaking, sensitive data disclosure, system prompt leakage
- Adversarial machine learning: privacy attacks, model extraction, transfer/boundary/noise attacks
- Data and training pipeline attacks: data poisoning, backdoor/trojan insertion, supply chain threats
- Agentic AI and model-to-model attacks: cross-LLM vectors, orchestration layer exploitation
- AI security testing, red team frameworks, MLOps/DevSecOps integration, and hardening best practices
- AI incident response, forensics, and capstone full-scope red team engagement