Governance of Automated Decisions: Audit, Oversight and Control for AI Decision Models Training Course

Transforming Decision-Making through Models

• Shifting from static rules and human judgement to scored options
• The pace and volume of automated actions
• Emerging failure modes and gaps in accountability

Inventorying Decisions and Classifying Risk

• Creating a catalogue of automated decisions
• Assessing impact, reversibility, and affected stakeholders
• Defining risk tiers and corresponding control requirements
• Assigning ownership to specific decision classes

Establishing Accountability and Human Oversight

• Distinguishing between meaningful and nominal human review
• Treating autonomy thresholds as critical governance decisions
• Defining escalation paths and response time expectations
• Identifying accountable owners and final sign-off responsibilities

Maintaining Decision Records and Explainability

• Essential components of a defensible decision record
• Persisting data on confidence, options, and context
• Ensuring traceability of models and versions
• Communicating decision logic to customers or regulatory bodies

Managing Policy and Boundary Controls

• Approving option sets and implementing change management
• Governing thresholds and conducting periodic reviews
• Defining prohibited decisions and mandatory hard stops
• Integrating policy logic alongside model outputs

Meeting Regulatory Expectations

• Obligations under the EU AI Act for decision-making systems
• Sector-specific regulations in finance, insurance, and healthcare
• Data protection standards and rights regarding automated decisions
• Maintaining documentation and evidence of conformity

Managing Vendor, Licensing, and Provenance Risks

• Comparing hosted decision APIs with open-weight alternatives
• Navigating licence terms for derivative and embedded models
• Mitigating supplier dependency and risks of service discontinuation
• Ensuring weight integrity and version pinning

Monitoring Bias, Fairness, and Outcomes

• Tracking outcome disparities across different groups
• Identifying proxy variables and biases in option framing
• Analysing signals from appeals and complaints
• Conducting regular fairness assessments

Responding to Incorrect Decisions

• Detecting patterns of systematic decision errors
• Implementing containment measures and suspending autonomy
• Executing remediation and providing customer redress
• Investigating root causes and updating controls

Conducting Assurance Reviews of the Decision Layer

• Identifying evidence required by auditors
• Validating claims regarding accuracy and calibration
• Sampling decisions and manual overrides
• Reporting findings and insights to the board