AI Security Foundation Certification

A training in support of ‘De Warmste Week’

All proceeds from this training go to De Warmste Week, the annual Flemish solidarity campaign. Data Protection Institute and Toreon are organising this training together, and every registration contributes.

A double win: you invest in your own expertise and support a good cause at the same time.

De Warmste Week shares the proceeds across all the projects it supports this year, under the theme “A warm home for everyone“.

One of them we’re glad to highlight: Knuffies.
Its volunteers bring warmth, comfort and closeness to babies and young children in hospital when their parents can’t always be there. Knuffies is an initiative of the non-profit Pleegouderforum and plans to expand to more hospitals in 2027.

After the training, we’ll let you know how much we contributed together to De Warmste Week.

AI is running in production at your organisation. Do you know how to secure it?

LLMs, RAG applications and AI agents are now part of your IT environment. They bring an attack surface that traditional application security doesn’t cover: prompt injection, data poisoning, exfiltration via tool calls.

In one day, you’ll learn how these systems are built, where the risks arise and which controls work. No machine learning theory, just the practice of deploying and managing AI. You’ll do hands-on exercises in every module and finish with an exam and a certificate.

This training was developed by Toreon.

Why take this training

• You gain a shared language for AI security: from LLMs and RAG to agents and MCP.
• You work with the frameworks the industry uses: OWASP LLM Top 10, OWASP Agentic AI Top 10, MITRE ATLAS, STRIDE, the EU AI Act and NIST AI RMF.
• You practise hands-on: a prompt injection lab, threat hunts on real architectures and a threat model as the capstone exercise.
• You learn from real incidents: timeline, root cause, detection and recovery.
• You earn a certificate with an exam, which you keep current each year with a 1-hour update.

Target audience

This training is for you if you deploy, build, secure or assess AI systems.

Learning objectives

After this day, you’ll be able to:

• name the building blocks of LLM, RAG and agentic architectures and point out where untrusted data comes in;
• map the attack surface of an AI application: supply chain, data, infrastructure and operations;
• recognise, prioritise and mitigate the risks in the OWASP LLM Top 10 and the OWASP Agentic AI Top 10;
• build a threat model for an AI application using DFDs, trust boundaries, STRIDE and MITRE ATLAS;
• turn lessons from real AI incidents into your own detection and response;
• link the essentials of the EU AI Act and NIST AI RMF to concrete controls in your development process (AI-SSDLC).

Teaching approach

• Classroom training.
• Technical and practice-oriented: every module includes an exercise or guided discussion.
• Mixed groups: every group is a mix of participants with different backgrounds and experience.
• Every module ends with what you’ll do with it in your role tomorrow.
• You take home a one-page “AI security operating checklist” for your role.

Outcome

A certificate that stays current.

After the training, you take the AISFC exam: 40 multiple-choice questions in 60 minutes, online proctored via APMG. You schedule the exam yourself when you’re ready. The pass mark is 60% (24 out of 40) and you get your result straight away, with a certificate and digital badge. One resit within 30 days is included.

How to prepare

You’ll need a basic understanding of AI, but no background in machine learning. We’ll provide you with preparatory materials.