Advanced AI Red Teaming (AI-300) is OffSec's specialized training for offensive security professionals targeting artificial intelligence ecosystems. The course moves beyond generic prompt engineering into technical exploits against LLMs, machine learning pipelines, and the underlying cloud infrastructure supporting AI deployments.
Key Features
- Hands-on labs focusing on RAG pipeline poisoning and vector database manipulation.
- In-depth analysis of multi-agent AI systems and agent-to-agent (A2A) protocol exploitation.
- Technical training on embedding inversion and information extraction from model weights.
- Exposure to AI supply chain vulnerabilities, including datasets, adapters, and malicious model artifacts.
- 24-hour practical proctored exam simulating a real-world enterprise AI environment.
Use Cases
- Assessing the security of generative AI applications and autonomous agents in production.
- Evaluating trust boundaries in complex multi-agent orchestration frameworks.
- Executing red team operations against AI-enabled cloud infrastructure and model servers.
- Identifying and mitigating risks in the machine learning supply chain and data retrieval layers.
