Agentic AI Risk and Security
About This Course
Agentic artificial intelligence (AI) is changing how organisations design, deploy, and govern digital systems. Unlike conventional applications, AI agents can call tools, access data, interact with other systems, and take multi-step actions on behalf of users. These capabilities create new security, data governance, vendor, operational, and incident response risks that existing application security and model risk frameworks may not fully address.
This course is designed for risk, assurance, compliance, cybersecurity, technology governance, and internal audit professionals who need to extend existing control frameworks to agentic AI systems. Participants will use an AI Agent Evaluation Framework to identify agent-specific risks, map them to preventive, detective, and corrective controls, and develop risk-register entries that can be adapted for workplace use.
The programme closes with a facilitated tabletop simulation of an agentic AI incident. Participants will practise escalation, containment, rollback, control remediation, and lessons-learned reporting, helping them return to work with practical tools for reviewing agentic AI deployments responsibly.
What You'll Learn
- Distinguish agentic risk from conventional application and model risk
- Assess model and data security exposure specific to agentic deployments
- Apply least-privilege design principles to agent tool permissions
Vendor Risk and Control Mapping
- Evaluate third-party and vendor artificial intelligence (AI) risk using a structured due diligence approach
- Map agentic AI risks to preventive, detective and corrective control types
- Build or update a risk register entry for an agentic AI initiative
Incident Response for Autonomous Systems
- Respond to a live agent incident using a structured escalation and containment process
- Distinguish containment and rollback options specific to autonomous systems
- Translate incident learnings into remediated controls
Entry Requirements
- Participants should have at least 1–2 years of experience in risk management, assurance, compliance, cybersecurity, technology governance, internal audit, data governance, vendor management, or related functions.
- Prior coding experience is not required. Participants should have basic familiarity with generative artificial intelligence (AI) concepts and organisational risk or control frameworks.
- This course is not intended for participants seeking hands-on AI model development or software engineering training.