Safe & Secure Agentic AI: Governance, Identity, and Red-Teaming Training Course

Foundations: Threat Modelling for Agentic AI

• Categories of agentic threats: misuse, privilege escalation, data leakage, and supply-chain risks.
• Adversary profiles and attacker capabilities specifically relevant to autonomous agents.
• Mapping assets, trust boundaries, and critical control points for agent operations.

Governance, Policy, and Risk Management

• Governance frameworks for agentic systems, including roles, responsibilities, and approval gates.
• Policy design covering acceptable use, escalation rules, data handling, and auditability.
• Compliance considerations and strategies for evidence collection during audits.

Non-Human Identity & Authentication for Agents

• Creating identities for agents using service accounts, JWTs, and short-lived credentials.
• Least-privilege access patterns and just-in-time credentialing methods.
• Strategies for identity lifecycle management, rotation, delegation, and revocation.

Access Controls, Secrets, and Data Protection

• Fine-grained access control models and capability-based patterns for agents.
• Secrets management, encryption in-transit and at-rest, and data minimisation practices.
• Safeguarding sensitive knowledge sources and PII from unauthorised agent access.

Observability, Auditing, and Incident Response

• Designing telemetry for agent behaviour, including intent tracing, command logs, and provenance.
• SIEM integration, alerting threshold configuration, and forensic readiness.
• Developing runbooks and playbooks for handling agent-related incidents and containment.

Red-Teaming Agentic Systems

• Planning red-team exercises: defining scope, rules of engagement, and safe failover procedures.
• Adversarial techniques such as prompt injection, tool misuse, chain-of-thought manipulation, and API abuse.
• Executing controlled attacks to measure exposure and impact.

Hardening and Mitigations

• Engineering controls including response throttling, capability gating, and sandboxing.
• Policy and orchestration controls such as approval flows, human-in-the-loop mechanisms, and governance hooks.
• Model and prompt-level defences: input validation, canonicalisation, and output filtering.

Operationalising Safe Agent Deployments

• Deployment patterns for agents, including staging, canary, and progressive rollout strategies.
• Change control, testing pipelines, and pre-deployment safety checks.
• Cross-functional governance involving security, legal, product, and operations playbooks.

Capstone: Red-Team / Blue-Team Exercise

• Execute a simulated red-team attack against a sandboxed agent environment.
• Defend, detect, and remediate as the blue team, utilising established controls and telemetry.
• Present findings, remediation plans, and proposed policy updates.

Summary and Next Steps