IT training across Europe - tecnovy makes you future-ready.

Switch to English
E-Learning

tecnovy AI Suite

AI Security & LLM Security

A weakness in an AI system is proven when an attack triggers it in the lab. If the same attack then fails, that establishes only the outcome of this test. By that rule you attack models, LLM applications and agents in a controlled lab environment with prompt injection and jailbreaks, provoke data leakage, document every finding and derive safeguards for development and operation, each one tested against the attack it is meant to stop.
Attendance certificate4.8/5 on ProvenExpert

What this module delivers.

Attack, harden, attack again: in the tecnovy AI Security & LLM Security training you test models, LLM applications and agents in that cycle. Your team then bases the approval of AI applications on documented test results.

How the training runs

Method
Attack and defence exercises in a controlled lab environment
Basis
tecnovy curriculum referencing OWASP and MITRE ATLAS
Outcome
tecnovy certificate of participation, no examination

Dates & booking

Choose a date that fits

No public dates are scheduled yet.

New dates are published regularly. Join the waitlist or plan an in-house course for your team.

Request in-house training

Fit

Who this module is designed for

Typical roles

  • You approve AI applications from an IT security standpoint, or hold them back.
  • You run penetration tests and now need to cover AI systems as well.
  • You build AI applications or design their architecture and want to know where they can be attacked.

Prerequisites

No prerequisites

You can start right away. What you need: basic knowledge of IT security, experience with LLM applications.

Consider instead Secure AI Architecture For those who want to plan security into the design of an AI system rather than attack finished ones.

Curriculum

AI Security & LLM Security Course in Detail

Five parts that lead from an overview of the attacks to the defence. After the threat landscape you attack LLM applications and agents in the lab environment, practise red teaming and finally derive safeguards whose effect you test against the same attacks.

01Threat landscape

This part maps the points at which an AI system can be attacked.

  • Attacks on data, such as poisoned training data
  • Attacks on models, such as their manipulation
  • Attacks on LLM applications
  • Attacks on agents and their tools
02Attacks on LLM applications

Here you carry out the attacks yourself in the lab environment.

  • Direct prompt injection: instructions in your own input
  • Indirect prompt injection: instructions in websites or files the model processes
  • Jailbreaks that make a model ignore its safety rules
  • Data leakage through the model's answers
03Agents and tools

This part shows how far an attack reaches once a model is allowed to act.

  • Excessive permissions of agents (Excessive Agency)
  • Insecure tools and how they are called
  • Risks through MCP servers; the Model Context Protocol (MCP) connects tools and data sources
04Red teaming

You attack an AI system the way an attacker would and record what you find.

  • Introductory red-teaming exercises against models, applications and agents
  • Documenting each weakness with the attack that proves it
  • All attacks only in the controlled lab environment
05Safeguards

Finally you derive safeguards from your findings and test them against the same attack.

  • Guardrails: rules and filters for a model's inputs and outputs
  • Limiting the permissions of agents and tools
  • Monitoring in operation
  • Incident response, the planned reaction to a security incident

Outcome

What you will be able to do afterwards

  1. 01

    You classify attacks by whether they target data, models, applications or agents.

  2. 02

    You run direct and indirect prompt injection against an LLM application and show its effect.

  3. 03

    You test whether jailbreaks override a model's safety rules and whether confidential data leaks out.

  4. 04

    You check agents for excessive permissions, insecure tools and risks that arrive through MCP servers.

  5. 05

    You document every weakness together with the attack that proves it.

  6. 06

    You derive guardrails and permission limits from your findings and test them against the same attack.

  7. 07

    You plan how monitoring detects an attack in operation and how your team responds to it.

Why tecnovy

What you get on top with us

01

No Slideshow, Hands-On!

Promised: no PowerPoint marathon. We work in groups, tie theory to practice, and you get real project examples from our experienced trainers plus the exchange with like-minded people.

02

Certificate Showroom

Get your certificate of participation and, if you have one, add your exam certificate from E-Learning. Fully automated, beautifully designed. Just for you, only at tecnovy.

03

Flexible Date Change

If you are not able to attend the course, you can rebook your training free of charge up to one week before the start of the training.

04

Attend Twice, Pay Once

You are welcome to attend the training online again within a year as a refresher.

05

Learn from Experts

We always guarantee you the use of didactically and methodically first-class qualified trainers who draw their knowledge from training experience as well as professional practical and project experience.

FAQs

Frequently asked questions

01May I try the attacks from the training on real systems?
Only on systems whose owner has explicitly permitted you to test them. In the tecnovy AI Security & LLM Security training you attack only systems in a controlled lab environment. Unauthorised attacks on other people's systems can be a criminal offence, in Germany for example under sections 202a et seq. of the Criminal Code (StGB).
02Will our AI application be secure after the training?
No training makes an application secure, and neither does a passed test. In the tecnovy AI Security & LLM Security training you learn to attack an application deliberately, document what you find and test safeguards against the same attack. A test proves which attacks failed, not that no others exist.
03What does red teaming mean for AI systems?
Red teaming means attacking an AI system the way a real attacker would, so you find its weaknesses first. In the tecnovy AI Security & LLM Security training you run introductory red-teaming exercises against models, applications and agents in the lab. You record each weakness together with the attack that triggers it.
04What role do OWASP and MITRE ATLAS play in the training?
Both are reference points of the curriculum in the tecnovy AI Security & LLM Security training, and no examination by OWASP or MITRE is attached to it. The 2026 OWASP Top 10 for LLM Applications lists risks such as Prompt Injection and Excessive Agency. MITRE describes ATLAS as a knowledge base of attacks on AI systems, built on observed attacks and demonstrations by red teams (as of November 2026).
05How does AI Security & LLM Security differ from Secure AI Architecture?
AI Security & LLM Security tests AI systems by attacking them, while Secure AI Architecture plans their security before they are built. In the tecnovy AI Security & LLM Security training you look for weaknesses in the lab and derive safeguards from them. If you want to design the trust boundaries, rights and isolation of a new system, Secure AI Architecture is the better fit.
06Do I also need the OWASP Top 10 for LLM and GenAI Applications training?
No, the two tecnovy trainings stand on their own. OWASP Top 10 for LLM and GenAI Applications works through the ten risks on the list one by one with examples and countermeasures. In the tecnovy AI Security & LLM Security training you carry out the attacks yourself, including on agents and their tools, and document what you find.
07Do I need to be a penetration tester?
No, experience with penetration testing is not required. For the tecnovy AI Security & LLM Security training, basic IT security knowledge and experience with LLM applications are enough, and there are no formal admission requirements.
08Does the training end with an examination?
No, the tecnovy AI Security & LLM Security training ends without an examination and leads to no certification. With at least 80% attendance you receive the tecnovy certificate of participation.

What does your training at tecnovy look like?

AI Security & LLM Security

No dates yet

We will let you know as soon as the next date is scheduled.

Feedback Form
What should we improve?

Tell us what we can do better next time.

Success
Error

© Copyright 2026, tecnovy GmbH, All Rights Reserved