[VIDEO] Research Seminar on Trustworthy AI in the Banking and Financial Sector

28/05/2026

A video recap of the FinAI-Lab research seminar on trustworthy AI!

A look back at the first research seminar on trustworthy AI, organized as part of the Fin AI Lab—a joint research lab between BNP Paribas and Télécom Paris—co-directed by Albert Bifet, Lea Déléris, Mariam BARRY, and Thomas Le Goff.

Held on May 28, this event brought together researchers and experts to discuss a topic at the heart of current issues: agent-based AI systems. The presentations sparked rich and constructive discussions on the technological, legal, and societal challenges associated with these new systems.

Governing Agentic AI under EU Law

Thomas Le Goff, associate professor of law and digital regulation at Télécom Paris, and Nina Lasbleiz, a research engineer at Télécom Paris, presented:

  • Recent technological developments in the field of agent-based AI;
  • The legal issues raised by the development of these systems;
  • The challenges associated with compliance with the GDPR and the AI Act.

“Agent-based AI raises questions about the implementation of certain regulatory obligations, particularly with regard to human oversight, liability, and data protection.”
— Thomas Le Goff

“Deploying agent-based AI systems without safeguards inevitably leads to harmful behavior and violations of the legal framework.”
— Nina Lasbleiz

Trust, Trustworthiness and Game Theory in LLM-Based Multi-Agent Systems

Azzedine Idir Ait Said, a research engineer at Télécom Paris, has explored trust mechanisms in multi-agent systems based on large language models (LLMs):

  • Game theory as a framework for analyzing strategies, incentives, and mechanisms of cooperation;
  • Trust as an emergent property of interactions between agents;
  • Approaches for evaluating, auditing, and designing more reliable, interpretable, and controllable systems.

In traditional software systems, trust relies heavily on explicit metrics, transparent methods, and deterministic verification mechanisms. With generative AI, these benchmarks still exist, but they become much more difficult to establish.”
— Azzedine Idir Ait Said

This first seminar exemplifies the Fin AI Lab’s mission: to advance cutting-edge research on trustworthy AI by bringing together scientific, technological, and regulatory expertise to address the challenges of the future.

The speakers’ presentations are available via the following links:

  1. Introduction: https://my.mediahub.bnpparibas/AssetLink/x3k1x2n2er4cp5e0717784lcgx2d51x3.mp4
  2. Governing Agentic AI under EU Law: https://my.mediahub.bnpparibas/AssetLink/avf580a8k145542i5tiaut3e0aly83i5.mp4
  3. Trust, Trustworthiness and Game Theory in LLM-Based Multi-Agent Systems: https://my.mediahub.bnpparibas/AssetLink/j2gmwo1612frpuxib5q768mwl0nea27u.mp4
  4. Conclusion: https://my.mediahub.bnpparibas/AssetLink/6p20va5v672ad7g7mgm64px5g4016a6e.mp4