The Bank of France has published an assessment of artificial intelligence (AI) in central banking and financial supervision, reviewing operational gains from existing applications and the safeguards needed for wider deployment. AI supports data processing, economic analysis and supervisory work, but its use requires human oversight and controls over model reliability and confidential information.
The bank rolled out five priority applications in 2025 and early 2026, including a conversational assistant for supervisors. Initial user feedback indicated that this tool saved around two hours per staff member each week. An application detecting unusual transactions reduced false alerts in work against money laundering and terrorist financing. Other deployments support human resources enquiries, staff handling household applications, and analysis of structured financial products.
The assessment identifies inaccurate outputs, bias and model drift as risks requiring ongoing controls. An internal AI Advisory Group established in 2025 has issued around 20 opinions on new and existing applications. France’s Prudential Supervision and Resolution Authority (ACPR) is also preparing for its role in implementing the European Union’s AI Act through work on legal questions, AI auditing and regulatory coordination. The bank’s 2026-28 strategic plan focuses on transforming business processes and making AI accessible to all staff.