HumenhukAIGP Study Guide
Module 7: Governing AI Deployment, BoK I.C

Adapting existing policies for AI

Reviewed by Victor Humenhuk. Published . Reviewed .

Review the current policy framework for gaps first, then tailor what exists and add what's missing across five areas - data privacy, security, intellectual property, engineering/model ops and open source. Adopt a risk-centric approach, evaluate frontier intent, integrate holistically and address AI procurement explicitly.

Review the current policy framework for gaps first → tailor what exists, add what's missing, and start from a solid data governance framework if you have one.

  • Data privacy → compliance with regulation while safeguarding sensitive information processed by AI.
  • Security → update for AI-specific risks like adversarial attacks, vulnerabilities in ML models.
  • Intellectual property → ownership and usage rights for AI-generated outputs and proprietary algorithms.
  • Engineering / model ops → the development life cycle and best practices for AI engineering.
  • Open source & platforms → the organisation's position on open-source models and platforms like AWS or Google's Model Garden.
  1. Adopt a risk-centric approach → resources are limited, so concentrate on the most significant, highest-risk areas.
  2. Evaluate cutting-edge intent → incorporate risk acceptance and watch legal requirements for automated decision-making and frontier models, the most advanced capabilities of AI.
  3. Integrate holistically → align AI governance with existing processes and make policies flexible across laws, industries and technologies.
  4. Address AI procurement → if models will be procured externally, that must be explicitly addressed in governance policies.

Key terms - quick answers

What is “Frontier models”?
The most advanced capabilities of AI, warranting heightened legal and risk attention.

Sources and study method

This independent lesson uses active recall, spaced retrieval and scenario practice. Read the full study method.