Advanced Training · DART Cybersecurity
Applied AI for Cybersecurity Professionals
A 5-day intensive programme equipping cybersecurity professionals with the knowledge and hands-on experience to understand, secure, evaluate, and respond to AI systems.
Artificial Intelligence is rapidly transforming how organisations operate, innovate, and defend themselves. From predictive analytics and automation to generative AI and agentic systems, AI technologies are increasingly embedded in business processes, critical infrastructure, and cybersecurity operations. As organisations accelerate AI adoption, cybersecurity professionals must develop a practical understanding of how these systems work, the risks they introduce, and the controls required to secure them.
AI systems differ fundamentally from traditional software. They rely on complex data pipelines, probabilistic models, and adaptive behaviour that can create new attack surfaces and security challenges. Large Language Models (LLMs), retrieval systems, autonomous agents, and AI-powered workflows introduce vulnerabilities that are not fully addressed by conventional cybersecurity practices. Threats such as prompt injection, model extraction, data poisoning, adversarial manipulation, and supply-chain compromise require defenders to understand both AI technologies and the security implications of their deployment.
At the same time, AI presents significant opportunities for cybersecurity teams. Security professionals are increasingly expected to evaluate AI solutions, participate in AI governance initiatives, define security requirements for AI deployments, and respond to incidents involving AI-enabled systems. To perform these responsibilities effectively, they need a practical understanding of AI architectures, attack methodologies, governance frameworks, and operational security controls.
This 5-day course equips cybersecurity professionals with the knowledge and hands-on experience needed to understand modern AI systems, identify and mitigate AI-related risks, evaluate AI deployments, develop governance and security requirements, and respond effectively to AI security incidents. Through a combination of lectures, practical exercises, and real-world case studies, participants will gain the skills required to support secure AI adoption within their organisations while strengthening their ability to defend against emerging AI-enabled threats.
Introduces the modern AI landscape: the four ML paradigms (supervised, unsupervised, reinforcement, generative) and where each is used in corporate and cybersecurity contexts; the end-to-end ML lifecycle from data ingestion and cleaning through deployment and monitoring; and the technology shift from traditional ML to Foundation Models and LLMs. Participants then study the building blocks of modern AI systems — LLM architectures, prompting and reasoning, memory and Retrieval-Augmented Generation (RAG), and tool integration via function calling and MCP — and the AI supply chain (developers, deployers, operators, users; models, data, frameworks, hardware, compute). The module is highly hands-on: learners progressively build a simplified agentic AI system in n8n (no coding required) to see first-hand how each lifecycle stage and component expands the attack surface.
Focuses on the evolving threat landscape, combining traditional software vulnerabilities with new attack vectors unique to ML and GenAI. Participants work through OWASP's Top 10 for LLM and GenAI applications, then deep-dive into prompt injection and jailbreaking, tool and RAG exploitation, supply-chain and model-loading attacks, and broader techniques catalogued in MITRE ATLAS. Each topic follows an Attack–Defend–Validate workflow on the agentic system built in Module 1. Real-world case studies ground the concepts in operational practice.
Equips participants with governance, risk, compliance, and threat-modeling approaches tailored to AI. Covers MITRE ATLAS for structured AI threat identification, the NIST AI Risk Management Framework for risks beyond security (bias, privacy, transparency, safety), and standard organisational procedures for AI integration — ethical considerations, data governance frameworks, AI Bills of Materials (AIBOM), model registries, and policies for employee use of external AI tools.
Extends incident response practice into AI-specific environments: detecting indicators of compromise, performing forensic analysis on prompt-injection traces, poisoned data, or malicious model behaviour, and implementing containment and remediation. Addresses unique challenges including non-deterministic outputs, distributed attack surfaces, and compromised training/retrieval pipelines. The day culminates in a team-based competitive exercise that also requires participants to brief technical and non-technical “stakeholders” (role-played by trainers) on their findings and decisions.