Strategic actions towards AI governance: Policy and structure

This sub-guideline is part of the guideline Strategic actions towards AI governance. Refer to the main guideline for context and an overview.

This sub-guideline provides guidance and recommendations for the design and development of AI governance policies and structures in parliaments.

When a parliament decides to adopt AI-based systems and services, it embarks on a journey that requires careful planning and a multidisciplinary approach. This journey begins with the recognition that AI governance is not just an IT issue: it is a matter that touches every aspect of parliamentary operations.

The first step is to assemble a diverse team. Executive boards, legal departments, and business and IT units all have crucial roles to play. This team will work together to create a governance structure that integrates business needs, legal and regulatory considerations, and technological insights. The exact nature of this structure will vary from parliament to parliament, reflecting each institution’s unique culture and existing working methods.

When it comes to establishing a governance structure, parliaments have two main options: either they add AI capacity to an existing board, or they create a dedicated AI-specific board. Both approaches have their merits, and the choice will depend on each parliament’s specific circumstances and resources.

Regardless of the chosen approach, the AI governance structure will have a wide-ranging set of responsibilities, including the following:

• Approving and monitoring AI policy and strategy
• Overseeing the development of an AI code of ethics
• Managing budgets for AI research and development
• Supervising AI projects from initiation to completion
• Dealing with ethical issues as they arise
• Monitoring AI systems throughout their life cycle, including defining criteria for when these systems should be decommissioned

Within this framework, it is crucial to define key bodies. These typically include the following:

• A high-level “governing body” (which may be part of IT or corporate governance)
• A body responsible for IT and data science
• Business units that act as stakeholders in the AI system life cycle
• An ethics committee – either a new entity or an adapted existing data committee – to manage the unique ethical challenges posed by AI

While a central group should oversee these roles, day-to-day responsibilities can be distributed across various areas of parliament. This might involve assigning tasks to existing functional areas or creating new units specifically to manage AI-related work. This approach ensures comprehensive governance while allowing for flexibility in implementation.

By taking these steps, parliaments can create a robust governance structure that enables them to harness the benefits of AI while effectively managing its risks and ethical implications.

Actions:

• Assemble a multidisciplinary team from across parliament to lead AI governance efforts.
• Choose and implement either an integrated or a dedicated AI governance board structure.
• Define and assign key roles and responsibilities for AI governance, including policy approval, ethical oversight and project management.
• Establish or adapt an ethics committee to address AI-specific ethical challenges.
• Develop a clear AI life cycle management process, from project initiation to system decommissioning.

The purpose of a parliamentary AI governance policy is to establish a unified approach to AI use within the parliamentary environment. This policy will set measurable standards and provide top-level monitoring for AI implementation.

The policy development process involves creating a dedicated working group, defining objectives, and establishing ethical principles aligned with national regulations and international best practices. It must address legal compliance, data protection and digital public service legislation.

The policy will outline roles and responsibilities for all stakeholders involved in the AI life cycle, from data scientists to legal experts. It will also define AI-supported business processes, prioritizing them while considering risk tolerance and regulatory requirements. This includes specifying conditions for AI use, prohibited areas, and approval processes for certain AI applications.

Guidelines for AI sourcing, development and outsourcing will be established, along with clear communication strategies to inform all parliamentary staff and MPs about the policy.

Regarding generative AI, the policy will provide clear usage guidelines and detail necessary precautions. While recognizing potential risks, it will also encourage innovative experimentation in a controlled manner, avoiding outright bans that might lead to unauthorized use on personal devices.

Actions:

• Establish an AI policy working group to lead the development process.
• Outline roles, responsibilities and processes for AI governance and implementation.
• Create guidelines for AI use, including prohibited areas and approval processes.
• Define mechanisms for ensuring regulatory compliance.
• Develop a communication strategy to inform all staff and MPs about the AI policy.

The Guidelines for AI in parliaments are published by the IPU in collaboration with the Parliamentary Data Science Hub in the IPU’s Centre for Innovation in Parliament. This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International licence. It may be freely shared and reused with acknowledgement of the IPU. For more information about the IPU’s work on artificial intelligence, please visit www.ipu.org/AI or contact [email protected].