As the UN meets to adopt the Pact for the Future and the Global Digital Compact, effective and interoperable good governance of AI to minimize risks and maximize benefits will be a key issue. As governing bodies worldwide establish new AI frameworks, institutions and policies, we risk creating a fragmented policy landscape for a technology that knows no borders. How do policies and institutions align with one another to foster trust and transparency? To what degree are these policies or frameworks interoperable? How are issues of civil and human rights considered in different AI frameworks? How can companies ensure compliance across geographies?

Foundation models continue to be a focus for policymakers, with the EU’s AI Office developing Codes of Practice for general purpose model providers. However, as the AI ecosystem grows more complex, with more actors and shifting roles, it’s crucial to understand this intricate web to craft effective policy. The integration of advanced AI assistants or “agents” into our daily lives represents a significant shift, potentially altering how we interact with technology and raising new governance challenges. How can we anticipate and prepare for the shifts in the AI value chain that are likely to occur in the coming months and years? What are the roles and responsibilities of actors across the value chain, from model providers to hosting services and application developers, and how should these be codified in policy including at the global level?

Developing meaningful AI policies requires inclusive, multistakeholder participation. How can policymakers and tech companies ethically engage stakeholders on AI issues? How do you incorporate inclusive AI participation policies within global frameworks? What are the best practices for community engagement, particularly for historically marginalized and underserved communities that could be reflected in global policy? How do you ensure the process is participatory and not extractive? Members of PAI’s Global Task Force for Inclusive AI discuss their advice for policymakers based on the recently published Recommendations for Inclusive AI.