Curriculum

AI Governance · Regulatory Frameworks

Model Governance & Compliance

AI governance frameworks—EU AI Act, NIST AI RMF, ISO 42001—define risk classifications, documentation requirements, and audit standards for responsible AI deployment.

The idea

Regulatory pressure on AI is accelerating. The EU AI Act (2024) is the world's first binding AI regulation. It classifies AI systems by risk: unacceptable (banned: social scoring, real-time biometric surveillance), high-risk (strict requirements: medical devices, CV hiring, credit scoring), limited risk (transparency obligations), minimal risk (free to deploy).

High-risk systems must: maintain technical documentation, implement human oversight, ensure traceability through audit logs, and register in the EU database before deployment. Non-compliance fines: up to €30M or 6% of global annual revenue.

NIST AI RMF (US): a voluntary framework with four core functions—Govern, Map, Measure, Manage—for managing AI risk throughout the lifecycle. Complements SOC 2 and ISO 42001 for enterprise AI governance.

Risk = Likelihood × Impact × (1 − Controls)

Standard risk quantification. Controls (technical and procedural) reduce effective risk.

What to hold onto

  • Audit LoggingEvery inference call in a high-risk system must be logged with timestamp, input hash, output, model version, and decision rationale. Immutable audit trails are non-negotiable for compliance.
  • Human-in-the-LoopHigh-risk systems require meaningful human oversight—not rubber-stamp review. Systems must be designed so operators can understand, monitor, and override AI decisions.
  • Model RegistryA centralised registry tracks model versions, training data provenance, evaluation metrics, and deployment history. Essential for regulated industries (finance, healthcare, insurance).

Live: EU AI Act Risk Tiering Classifier

Classify AI systems into regulatory tiers: Minimal, Limited, High-Risk, or Unacceptable under EU AI Act.

Tier: MINIMAL RISK

Quick answers

What is the EU AI Act?

The EU AI Act is the world's first comprehensive AI regulation, enacted in 2024. It classifies AI systems by risk level: unacceptable (banned), high-risk (subject to strict conformity assessment, documentation, and human oversight requirements), limited risk (transparency obligations), and minimal risk (unrestricted). Non-compliance carries fines up to €30M.

What are AI Audit Logs?

AI audit logs are immutable records of every AI system decision, capturing timestamps, input data hashes, outputs, model version IDs, and operator actions. They provide the accountability trail required by regulations like the EU AI Act for high-risk applications in healthcare, credit, and employment.