Skip to Content
View all events

Agentic AI and Cybersecurity Risks

Published by Pearson

Intermediate content levelIntermediate

Learn to design, secure, and optimise autonomous AI agents for real-world applications

  • Design and implement agentic AI systems, integrating LLMs, memory, planning, self-adaptation, and reinforcement learning for autonomous decision-making.
  • Analyze security risks in agentic AI systems, with a focus on identifying and countering adversarial attacks, model manipulation, and self-replicating threats.
  • Engage in hands-on exercises with multi-agent systems for secure, efficient task delegation and coordination in cybersecurity, finance, and robotics.

Agentic AI represents the next evolution of artificial intelligence, enabling models to autonomously perceive, reason, and act within dynamic environments. As AI systems transition from passive predictive models to agentic architectures capable of self-directed decision-making, significant security challenges arise, such as vulnerability to adversarial manipulation and self-replicating attacks.

This course is focused on the security implications of agentic AI, teaching participants how to build and secure autonomous systems while addressing emerging threats. Participants will gain hands-on experience in developing resilient agentic AI systems, integrating technologies such as reinforcement learning (RL), large language models (LLMs) with memory and planning, and multi-agent interactions. Throughout the course, students will participate in practical exercises to secure these systems, apply risk assessment frameworks, and develop strategies to mitigate vulnerabilities. By the end, participants will be equipped to design and deploy secure, efficient agentic AI systems across industries, including cybersecurity, finance, robotics, and autonomous decision-making.

What you’ll learn and how you can apply it

  • Apply core agentic AI principles, including decision-making loops, autonomy, and system security.
  • Distinguish agentic systems from passive AI, with applications in cybersecurity, automation, and decision-making.
  • Implement memory, planning, and self-improvement techniques, prioritizing security and system optimization.
  • Deploy agentic AI in enterprise environments, emphasizing secure automation, AI assistants, and robotic processes.
  • Design governance and ethical standards, ensuring compliance, security, and responsible AI use.

This live event is for you because...

  • You are an AI engineer, researcher, data scientist, or cybersecurity professional wanting to explore the next wave of AI autonomy.
  • You aim to build agentic AI systems with memory, planning, and decision-making capabilities.
  • You are responsible for governance, risk, and compliance of AI automation and assessing regulatory and ethical considerations.

Prerequisites

This course is designed for professionals with a basic understanding of AI and machine learning concepts. No prior experience with agentic AI or multi-agent systems is required.

Recommended prerequisites

  • A general understanding of LLMs, reinforcement learning, and adversarial AI is helpful but not required.
  • A willingness to engage in hands-on exercises and security assessments.
  • A computer with a stable internet connection is required for interactive sessions.

Course Set-up

  • Attendees will be required to access the platform using a computer with a stable internet connection. No specific software or setup is required beyond this, as all resources will be provided directly through the interactive platform.

Recommended Preparation

Recommended Follow-up

Schedule

The time frames are only estimates and may vary according to how the class is progressing.

Session 1: Introduction to Agentic AI and Cybersecurity Risks (50 minutes)

  • Overview of Agentic AI Architectures
  • Threat Landscape for Autonomous AI
  • Case Studies of Real-World AI Cyber Threats
  • AI Agent Attack Surface Analysis
  • Threat Modeling Exercise
  • AI Agent Takeover Simulation
  • Q&A Session

Break (10 minutes)

Session 2: Adversarial Manipulation and Risk Propagation in Agentic AI (50 minutes)

  • Categories of Attacks Against Agentic AI
  • Risk Propagation in Autonomous AI Ecosystems
  • Case Study: Attacks on AI-Powered Autonomous Cybersecurity Agents
  • Live Demonstration: Simulating an Adversarial Attack
  • Hands-on Risk Assessment
  • Prompt Injection & Hijacking
  • Q&A Session

Break (10 minutes)

Session 3: Red Teaming and Security Assessments for Agentic AI (50 minutes)

  • Principles of Red Teaming for Agentic AI
  • Security Assessment Frameworks for AI Agents
  • Introduction to Adversarial Robustness Testing for AI Agents
  • Red Teaming Exercise: Implementing an AI Security Assessment
  • Testing AI Agent Robustness
  • Simulating an AI Agent Manipulation Attack
  • Q&A Session

Break (10 minutes)

Session 4: Defending Agentic AI: Security Strategies and Governance (50 minutes)

  • Defensive AI Strategies
  • Ensuring Alignment and Safety in AI-Driven Decision-Making
  • Ethical and Legal Challenges in Autonomous AI Security
  • Future Risks: Autonomous AI-Driven Cyber Threats
  • Implementing Defenses
  • Red Team vs. Blue Team Exercise
  • Q&A Session

Course wrap-up (10 minutes)

Your Instructor

  • Dr. Petar Radanliev

    Dr. Petar Radanliev lectures and supervises postgraduate master’s students’ research dissertations on AI and cybersecurity at the Department of Computer Science, University of Oxford. He is also a Lecturer/Instructor at Pearson and O’Reilly (USA), while conducting research on digital identity system security at the Alan Turing Institute, based at the British Library in London. After completing his PhD in 2013/14, Petar held postdoctoral research appointments at Imperial College London, the University of Cambridge, the Massachusetts Institute of Technology, and the Department of Engineering Science at the University of Oxford, where he remained for seven years before moving to his current position. His work spans artificial intelligence, cybersecurity, post-quantum security, and blockchain security. This research has led to an H-index of 25 (as indexed by Web of Science and Scopus), over 3,700 citations, more than 100 peer-reviewed publications, and four authored books. In recognition of his contributions, Petar has received major funding awards, including a Fulbright Fellowship and the Prince of Wales Innovation Award.

Skill covered

AI Security