Course · NCAIS
NSD Certified AI Security Professional
Secure AI. Use AI to secure.
A course by the Cybersecurity & AI Initiative, an ISAC Foundation centre.
- Length
- 18 hours
- When
- Weekends: Saturday and Sunday, two hours each
- Format
- Live instruction, demonstrations and case exercises
- Certification
- NSD certification: a knowledge assessment and a case-based enterprise scenario
- Before you start
- NCBAI recommended
About the course
An AI security certification built around both sides of the equation. Security for AI: protecting LLMs, RAG, models, data, agents, identities, MCP and A2A, and the AI supply chain. AI for security: applying copilots and security agents to the SOC, investigation, detection, hunting, vulnerability prioritisation and response.
What you become
An AI security professional who can map the AI attack surface, design secure AI architecture and use AI safely in the SOC.
Who it is for
- SOC analysts and incident responders
- Security architects and engineers
- AppSec, DevSecOps and cloud security professionals
- Pentesters, red and blue teamers
- AI, ML and platform engineers
- Cybersecurity leaders and technology-risk professionals
Curriculum
The AI attack surface and LLM security
The modern AI stack, prompt injection, jailbreaks, OWASP LLM Top 10
The modern AI stack and its attack surfaces across model, data, application, agent and infrastructure; prompt injection, jailbreaks, context manipulation, data disclosure and guardrail bypass; OWASP GenAI LLM Top 10 and application-layer defences.
RAG, model and agentic AI security
RAG poisoning, model provenance, agents, MCP and A2A
RAG poisoning, malicious documents, vector stores, fine-tuning and model provenance; goal hijacking, memory poisoning, tool misuse and excessive autonomy; MCP, A2A, non-human identities and multi-agent trust boundaries.
Threat modelling, architecture and red teaming
OWASP Agentic Top 10, MITRE ATLAS, MAESTRO, PyRIT, garak
OWASP LLM and Agentic Top 10, MITRE ATLAS and CSA MAESTRO; agentic threat modelling; secure AI SDLC, LLMOps and AgentOps, AI gateways and runtime controls; AI red teaming with PyRIT, garak and promptfoo.
Monitoring, incident response and AI for security
AI telemetry, incident response, AI-assisted and agentic SOC
Prompt, model, RAG, agent and tool telemetry; anomaly detection and agent containment; AI incident response and forensics; AI-assisted investigation, hunting and detection engineering; the agentic SOC.
What is included
- Optional free practice: PortSwigger LLM labs, OWASP AI Vulnerabilities Playground, Microsoft AI Red Teaming Playground, PyRIT, NVIDIA garak
Enrol
Send an enquiry to the Cybersecurity & AI Initiative from this page.