Global AI Governance

"For the AI revolution to mean progress for everyone, it is essential to design a collective and inclusive governance architecture to address all the challenges of the technology.” Henri Verdier, Summit Envoy for Global Governance of IA.

The global governance framework for artificial intelligence is currently piecemeal. Some countries and regional organizations, like the European Union, have established their own regulatory frameworks autonomously. At international level, however, there is no unified governance, even if many actors are and must be involved, as artificial intelligence raises all sorts of questions, including safety, ethics, environmental impact, access, competition, cultural diversity, data protection, interoperability of standards, military use, integrity of information, open technical standards and actors’ sovereignty.

However, in the absence of coordination, multiple parallel initiatives have been launched in very different formats, like the G7 Hiroshima AI Process and the Bletchley Park Summit in 2023, and the Tech Accord to Combat Deceptive Use of AI in 2024 Elections that was signed by companies alone in Munich in February 2024.

This creates a complex landscape that is often insufficiently inclusive and where frameworks sometimes compete. In this context, the accountability of artificial intelligence actors is upheld very strictly in terms of the commitments demanded of them and at the same time very loosely due to the lack of competent control institutions. Meanwhile, various norms co-exist depending on the regions concerned, which penalizes smaller actors that are less well-equipped to navigate between legal systems.

Echoing the vision for more inclusive and effective governance of artificial intelligence promoted by the Global Digital Compact, which now needs to be established tangibly, the primary aim of the Summit in this area is to clarify and design a shared and effective governance framework with all relevant actors so that science, solutions and standards are better shared.

Inclusiveness is the watchword. Today, only seven countries worldwide are participating in the major international artificial intelligence initiatives, while 119 are totally absent from them. Moreover, private actors and civil society need to be fully included in this governance, which must not be restricted to public authorities alone.

As such, the chosen approach is a collective one, involving many consultations with various stakeholders from across the world. The aim is clear: to build a consensus on the global governance framework for artificial intelligence, with and for all parties.

With 70 partners, including States, international organizations, businesses and civil society representatives, the aim is to survey the artificial intelligence governance system’s needs, actors, fundamental principles, limits and room for progress.

One single governance initiative is not the answer. Rather, existing initiatives, like the Global Partnership on Artificial Intelligence (GPAI) that will meet in an expanded format on 9 February to agree on a new action plan for 2025, need to be coordinated to build a global, multi-stakeholder consensus around an inclusive and effective governance system for artificial intelligence.