AI Ethics and Compliance Training Course

Participants can perform a structured assessment of AI harms, controls and residual risk.

• Risk taxonomy covering individual, organisational, societal, security and environmental harms
• AI impact assessment scope, triggers, stakeholders, evidence and review dates
• Rights-based assessment of discrimination, privacy, autonomy and due-process impacts
• Scenario analysis for foreseeable misuse, failure modes and vulnerable groups
• Inherent-risk scoring using likelihood, impact, reversibility and scale criteria
• Residual-risk acceptance thresholds, treatment plans and documented approvals

Participants can identify data protection and information-governance controls across the AI lifecycle.

• GDPR principles applied to training data, prompts, outputs and model monitoring
• Lawful basis, purpose limitation and data minimisation for AI processing
• Data provenance records for source, consent, licensing, quality and restrictions
• Data protection impact assessment triggers for high-risk automated processing
• Retention schedules, deletion controls and access permissions for AI datasets
• Copyright, database rights and trade-secret risks in generative AI inputs

Participants can specify controls to detect discriminatory outcomes and support meaningful human intervention.

• Sources of bias in problem framing, labels, sampling and feature selection
• Protected characteristics, proxy variables and intersectional harm assessment
• Fairness metrics: demographic parity, equal opportunity and calibration trade-offs
• Disaggregated performance testing across relevant population and user segments
• Human-in-the-loop design: authority, competence, time and override mechanisms
• Appeal routes, contestability records and adverse-decision communication requirements

Participants can create documentation and communication controls suited to AI system risk.

• Transparency requirements for users, decision subjects, operators and oversight bodies
• Model cards documenting purpose, data, performance, limitations and intended users
• Datasheets for datasets covering collection methods, composition and known limitations
• Explainability methods: feature importance, counterfactual explanations and local explanations
• User notices distinguishing AI-generated content, automation and human review
• Technical documentation evidence for design choices, tests, changes and approvals

Participants can map applicable AI and data protection obligations to system controls.

• EU AI Act prohibited practices, high-risk systems and transparency duties
• Risk classification factors: intended purpose, deployment context and affected persons
• High-risk system obligations for risk management, data governance and logging
• General-purpose AI responsibilities, documentation duties and downstream information provision
• GDPR automated decision-making provisions, safeguards and data-subject rights
• Regulatory horizon scanning for sector rules, national laws and regulator guidance

Participants can design proportionate AI controls using recognised risk and management-system frameworks.

• NIST AI RMF Govern, Map, Measure and Manage functions
• ISO/IEC 42001 management-system objectives, policies, roles and continual improvement
• Control mapping from identified risks to preventive, detective and corrective controls
• Key risk indicators, thresholds, alert ownership and management reporting
• Control testing evidence: logs, approvals, test results and exception records
• Three-lines model responsibilities for management, assurance and independent audit

Participants can establish lifecycle controls from design through post-deployment monitoring.

• Stage-gate controls from use-case proposal through retirement and archival
• Requirements traceability linking business objectives, risks, controls and test cases
• Validation plans for accuracy, reliability, robustness, security and fairness
• Model drift, data drift and performance degradation monitoring thresholds
• Change management for retraining, prompt updates, model replacement and configuration changes
• Incident response playbooks for harmful outputs, failures and security events

Participants can assess supplier AI services and define proportionate contractual safeguards.

• Supplier due diligence covering model purpose, training data and evaluation evidence
• Shared-responsibility mapping for provider, integrator, customer and end-user obligations
• Contract clauses for audit rights, incident notification, data use and deletion
• Assessing vendor claims about accuracy, bias, explainability and security
• Service-level measures for availability, support, change notice and remediation
• Exit planning for data portability, model replacement and service discontinuation

Participants can communicate AI assurance findings and prepare a practical implementation plan.

• AI governance dashboard metrics for inventory coverage, risk status and incidents
• Assurance reporting for executives, audit committees, regulators and product teams
• Finding statements linking control gaps, consequences, root causes and recommendations
• Prioritisation matrix using legal exposure, customer impact, urgency and implementation effort
• Ninety-day implementation roadmap with owners, milestones, dependencies and evidence
• Personal application plan for transferring controls to current AI initiatives