BNP Paribas extends its partnership with Mistral AI for three more years.
Initiated in 2023 and formalized in 2024, this collaboration now extends to software, bespoke solutions, and co-developed research, with direct integration of Mistral AI engineers and data scientists into the bank’s teams.
The immediate challenge is of a practical nature: speed. Marc Camus, BNP Paribas’s Director of Information Systems, sounded the alarm: “There is a lot of noise on the market regarding Mythos and whether Mythos is accessible to banks, especially European banks,” he admitted during a joint press conference with Mistral.
Unveiled in April by Anthropic, the Mythos model was specifically designed to detect vulnerabilities in software systems at an unprecedented scale and speed.
“What changes the game is the speed with which we must remediate vulnerabilities and their scale. We are discovering a large number of them simultaneously,” explains Marc Camus, adding that the bank “is working very, very hard” to prepare for this.
Asked about this competition around “Mythos,” Corentin Petit, Global Head of Solutions at Mistral AI, brushed off the technical details, promising forthcoming announcements, while recalling the startup’s positioning: “We will optimize the performance indicators that matter to our clients in the sector,” explicitly targeting the requirements of highly regulated markets like banking.
From front office to compliance: AI deployed at scale
Beyond the IT trench warfare, the renewal of this partnership aims to industrialize generative AI at every level of the banking group, combining technological performance with the protection of sensitive data through solutions deployed on-premise.
Concrete applications are already a reality for tens of thousands of collaborators.
Sophie Heller, Director of Transformation for BNP Paribas’s Retail and Consumer division, specifies that Mistral AI models already power internal tools as well as client-facing virtual assistants in France and Belgium. AI is also at the heart of compliance processes within Fortis, the group’s Belgian subsidiary.
The same momentum is seen in the bank’s Corporate & Investment Banking (CIB) arm. According to Charles Holive, Head of AI in this division, deployments in operation or in advanced pilot phases now cover automated data extraction from documents, equity research, and rapid retrieval of internal knowledge.
