Earlier this month, France’s President Emmanuel Macron hosted a summit on artificial intelligence aimed at pushing forward global governance efforts (“Macron unveils €109bn AI investment plan”, Report, February 11).
The initiative seemed, at first sight, like a disaster. Unlike previous summits that narrowly focused on mitigating AI safety risks, such as advanced cyber threats, the French considered topics ranging from the future of work to public interest AI.
Unsurprisingly, the resulting joint leaders’ statement was vague and a far cry from the action promised. Both the US and the UK — often considered the first and third most important countries in AI respectively — refused to sign the document.
However, when considering the wider context, a less damning picture emerges. Under Donald Trump, the US has already marked a clear pivot towards deregulation of AI. At France’s AI Action Summit, JD Vance, US vice-president, made it clear that AI opportunities, and not safety, was their priority. Thus, even if France had stuck to a more narrowly focused AI safety agenda, it seems unlikely that it would have received US government backing.
However, meaningful progress on AI governance can take place beyond intergovernmental agreement. And this is where France’s summit excelled. While the UK and South Korea AI summits were small-scale, France convened thousands of AI stakeholders from governments, industry and academia. Exchanges between these stakeholders provide the foundation for institutional and technical co-operation that can support the good governance of AI, even in geopolitically challenging times.
The lesson of Paris is that targeted and incremental intergovernmental negotiation, combined with a broader stakeholder convening, is the path with the maximum impact.
Huw Roberts
Doctoral Researcher, University of Oxford’s Internet Institute, Oxford, UK
