#European Affairs- AI & International Governance

On Tuesday, January 28, 2025, the Sciences Po American Foundation hosted a discussion on AI and International Governance between leading experts in the field. Camille François is an Associate Professor at Columbia SIPA, specializing in the misuse of digital technologies, including cyber conflict and online harm. Pierre Noro is a lecturer at Sciences Po and the Learning Planet Institute, focusing on tech governance, digital ethics, and sustainability. Derek Leist drives the safety and governance roadmap for the Data and Model Factory at IBM Research as Senior Technical Program Manager, AI Governance Lead, and lectures on responsible AI at the MIT Sloan Executive Education Program. The conversation explored the geopolitical and ethical implications of artificial intelligence in a rapidly evolving global landscape.

The discussion opened by highlighting the rapid evolution of AI over the past several years, transitioning from narrow AI to general-purpose AI systems. This shift has introduced new challenges in governance, regulation, and ethical responsibility. Noro addressed the concept of AI governance, particularly in relation to the upcoming AI Action Summit in Paris. Unlike its predecessor, the AI Safety Summit of 2023, the new summit reflects an expanded scope beyond existential threats, incorporating discussions on AI’s social impact, labor markets, public action, and environmental concerns. François elaborated on the need to address short term AI-related issues such as misinformation, hate speech, and AI generated content which already affect people. 

The conversation proceeded to examine the AI regulatory landscape across different regions. The panelists analyzed contrasting approaches to AI governance between the U.S., EU, and China. The U.S. is beginning to shift towards a more deregulatory stance, as future policies remain uncertain with the change of administrations. Meanwhile, the EU AI Act introduces a structured, risk-based approach to mitigating AI risks, with an AI Office dedicated to regulation. China’s regulatory approach raises questions about state-aligned AI development, particularly with the introduction of DeepSeek.

A central debate emerged regarding open-source versus closed AI systems. Noro noted that some major corporations advocate for closed models, citing built-in safeguards and corporate economic interests. Open-source models, however, may offer greater transparency and accessibility, fostering collaborative, community-driven innovation. Open-source models may also enable a more informed choice of which AI models to use for specific downstream tasks and a clearer understanding of their potential biases and other shortcomings.

Beyond government and corporate interests, the role of civil society in AI governance was a key focus. Civil society organizations have begun stress-testing AI systems for bias and vulnerabilities, acting as watchdogs in an environment where both governments and corporations may prioritize model performance and efficiency over ethics. François stressed the importance of continued public engagement, ensuring that AI governance remains an inclusive and pluralistic endeavor.

The discussion ended with an exploration of AI’s future trajectory. As AI continues to advance, international governance structures must evolve to address both immediate and long-term challenges while encouraging responsible innovation. The AI Action Summit in Paris represents a critical step in shaping a global approach that balances innovation, regulation, and ethical responsibility. The role of policymakers, corporations, civil society, and academia will be crucial in defining AI’s future and ensuring it serves the public good.

In concluding remarks, Thibaut Jaulin, an academic advisor and lecturer at Sciences Po PSIA, presented a new Master’s program that combines Technology and Global Affairs. The multidisciplinary social science program focuses on the impacts of technology on geopolitics, economics, governments, and citizens. It is available as a two-year program, as a one-year program after five years of professional experience, and as a joint degree with IE University.