[ONLINE] When Artificial Intelligence Acts: Security and Trustworthiness of AI Agents

→ Europe/Prague
ZOOM (ONLINE)

ZOOM

ONLINE

Description

Annotation

This webinar introduces practical challenges in security and trustworthiness for modern AI systems and autonomous AI agents. Using real-world incidents, experiments, and research examples, participants will explore risks such as prompt injection, excessive permissions, unsafe tool use, insufficient safeguards, and unexpected agent behaviour. The webinar also presents practical principles for deploying and using AI agents in a safer, more controlled way, including least privilege, access control, monitoring, separation of trusted instructions from untrusted data, and human approval for critical actions.

Benefits for the attendees, what will they learn

  • Better understanding of the security risks introduced by AI agents.
  • Ability to recognise common weaknesses in AI-enabled workflows.
  • Practical guidance for reducing risks when adopting AI tools.
  • Awareness of real AI security incidents and current research.
  • Basic principles for designing more secure, controllable, and trustworthy AI systems.

 

Participants will learn how AI agents differ from traditional AI applications from a security perspective, how attacks such as prompt injection can influence agent behaviour, and why permissions and tool access are particularly important. They will also learn practical mitigation techniques, including least-privilege access, isolation of untrusted content, monitoring, validation, and human-in-the-loop controls for sensitive operations.

Level

Suitable for participants who already have a basic understanding of AI or LLMs but do not need previous cybersecurity expertise.

Language

English

Prerequisites

  • Basic understanding of artificial intelligence and large language models
  • General familiarity with how AI tools such as ChatGPT or AI assistants are used
  • No previous experience with AI security or cybersecurity required

Tutor

Dominika Regéciová is a Lecturer at IT4Innovations, focusing on the intersection of AI and security. Previously, she worked as a Senior Researcher at Avast. She holds a master's degree in Information Security from the Faculty of Information Technology at Brno University of Technology (FIT BUT). Her work is driven by a passion for bringing formal theory and advanced AI methods into practical, real-world applications with an emphasis on security and reliability.

 

 

The LUMI AI Factory Service Center project is co-funded by the EuroHPC JU and the Participating States of Finland, the Czech Republic, Poland, Estonia, Norway, and Denmark under Grant Agreement No. 1012342048. The project is co-funded by the Ministry of Education, Youth and Sports of the Czech Republic (MC2503).

This course was supported by the Ministry of Education, Youth and Sports of the Czech Republic through the e-INFRA CZ (ID:90254).

All presentations and educational materials of this course are provided under the Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0) license. 

Registration
[ONLINE] When Artificial Intelligence Acts: Security and Trustworthiness of AI Agents