Cybersecurity · AI Security
Secure AI and LLM Applications by Design
Understand the security risks that come with AI and large language model applications, and learn how to design, test and protect them using defensive, responsible practices.
- Threat modelling for AI and LLM applications
- Prompt injection and unsafe model interactions
- Protecting sensitive data in AI systems
- Testing and securing AI applications
Course overview
About This Course
AI Security looks at the new risks that appear when applications are built on large language models and other AI components, and at the practical controls that reduce them.
Organisations are adding AI assistants, retrieval-augmented search and AI agents to their products and internal tools. These systems take natural-language input, can reach sensitive data and may act through connected tools, which creates attack surfaces that traditional application security does not fully cover.
You will learn how AI applications are assembled, where their trust boundaries sit, how risks such as prompt injection and data leakage arise, and how to design, test and monitor AI systems more securely.
AI Security sits between our Cybersecurity and AI & Generative AI learning paths. It suits security professionals moving into AI and engineers who build AI applications.
Who it is for
Who Should Take This Course?
-
Security professionals
You already work in security and want to understand and assess AI-specific risks.
-
Developers and AI engineers
You build applications on LLMs and want to design them securely from the start.
-
Cybersecurity learners exploring AI
You have security fundamentals and want to move into an emerging specialism.
-
Technical leads and architects
You are responsible for AI features and need to understand their risks and controls.
Learning outcomes
What You Will Learn
By working through the course, you will build towards being able to:
Explain how LLM applications are built from prompts, models, retrieval and tools, and identify their trust boundaries.
Threat-model an AI application by identifying its assets, entry points and likely risks.
Describe prompt injection and jailbreak risks and the layered defences that reduce them.
Identify ways sensitive data can be exposed through AI systems and apply data-protection controls.
Assess the security of retrieval (RAG) pipelines and of tool or agent access.
Plan security testing and evaluation for an AI application.
Apply secure design principles, guardrails and monitoring to AI systems.
Curriculum
Course Curriculum
The course is organised into modules. Session order and depth may vary by training arrangement.
-
Module 1: How AI applications work: a security view
The building blocks of AI applications and where risk enters.
Topics covered include
Practical activity: Map the components and data flows of a sample AI assistant.
-
Module 2: Threat modelling AI systems
A structured way to find and prioritise risks in AI applications.
Topics covered include
Practical activity: Build a threat model for a sample retrieval-based chatbot.
-
Module 3: Prompt injection and unsafe model interactions
Why natural-language input is hard to secure, and how to reduce the risk.
Topics covered include
Practical activity: Review example interactions and recommend mitigations for each weakness.
-
Module 4: Data protection and privacy
Keeping sensitive information safe in prompts, context and outputs.
Topics covered include
Practical activity: Identify sensitive-data risks in a sample application design and propose controls.
-
Module 5: Retrieval, tools and agent security
Securing what an AI application can read and what it can do.
Topics covered include
Practical activity: Review the permissions of a sample AI agent and reduce them to what it needs.
-
Module 6: Testing, controls and monitoring
How to evaluate AI applications and keep them secure over time.
Topics covered include
Practical activity: Write a security test plan for a sample AI application.
Practical learning
Hands-On Learning
Practical work focuses on designing, reviewing and testing AI applications defensively. Examples include:
The focus is defensive. Any testing exercises use purpose-built practice applications that you are authorised to test. Ask us which activities are included in your training arrangement.
- Threat-modelling exercises on sample AI application designs
- Reviewing example interactions for prompt injection weaknesses
- Data-protection reviews of retrieval and logging designs
- Writing security test plans and recommended controls
Before you start
Prerequisites and Learning Requirements
-
Recommended background
A basic understanding of security concepts and of how AI or LLM applications work is recommended. Cybersecurity Fundamentals and Generative AI Development provide useful background.
-
Technical prerequisites
Familiarity with how web applications and APIs work is helpful.
-
Software and tools
A computer with a reliable internet connection for live online sessions. Any accounts or tools used in practical activities will be confirmed when you enquire.
-
Prior experience
Not intended as a first course for complete beginners in either security or AI.
Learning Format
IT Champion Hub offers live, instructor-led online training and workshops. Formats, duration and schedules vary by course and are not the same for every program.
- Live online sessions led by an instructor
- Practical activities alongside explanations
- Upcoming batches and schedules shared on enquiry
Enquire to find out the formats and upcoming batches available for this course. Enquire about this course
FAQ
Frequently Asked Questions
Do I need a security or an AI background?
Some background in at least one of the two is recommended. Security professionals and AI developers both take this course, and the early modules cover the essentials of the other area.
Is this course about attacking AI systems?
No. The focus is defensive: understanding risks so that you can design, test and protect AI applications. Any testing exercises use practice applications you are authorised to test.
Why is AI Security listed under both Cybersecurity and AI & Generative AI?
AI Security sits between the two disciplines, so it appears in both sections of the Training Catalog. It is the same course.
What tools or software might I need?
A computer with a reliable internet connection for live sessions. Any accounts or tools used in practical activities will be confirmed when you enquire.
How can I enquire about upcoming training?
Select Enquire About This Course. The enquiry form will note that you are asking about AI Security, and we'll get back to you with upcoming batches and available schedules.
Ready to Explore This Learning Path?
Have questions about the curriculum, prerequisites, or training arrangements? Get in touch with IT Champion Hub to discuss your learning goals.