Societe Generale expects to reduce costs by $500 million to $600 million by 2029 through the increased use of artificial intelligence, with approximately $350 million of these savings already embedded. This forms part of the bank's broader strategy to achieve a cost base below EUR 16.3 billion by 2029, a 2% decrease compared to 2026 estimates. The French bank is actively pursuing cost reduction through operational efficiency, technological transformation, and AI, including a strategic agreement with Anthropic to accelerate corporate AI adoption. The bank aims for an annual revenue growth of about 3% and a return on tangible capital between 13% and 14% by 2029.
Despite the significant financial potential of AI, bank CEOs' enthusiasm is tempered by warnings from regulators. The European Banking Authority (EBA) emphasizes the need for human oversight in AI-driven processes, particularly in critical areas like credit assessments. Regulators like the EBA are developing playbooks for national supervisors to mitigate AI risks, especially for complex AI models and reliance on third-party vendors. While technology exists to fully automate tasks such as Know Your Customer (KYC) checks and fraud detection, regulators insist on maintaining a minimum level of "human in the loop" to address concerns like false positives and ensure explainability to customers and authorities.
The regulatory landscape for AI in finance is rapidly evolving. The EU's AI Act, passed in 2024, provides significant guardrails, but the technology continues to advance quickly. Experts like Bhavi Mehta from Bain & Co. caution banks about potential liability if AI outputs are incorrect or lack verifiability. Herbert Swaniker of Clifford Chance highlights the increased importance of transparency, audit rights, and performance assurance when AI is integrated into banking operations. The EBA's focus on "explainability" ensures that human bankers can always articulate how AI is used and how outcomes are reached.
Banks also face challenges beyond regulation. While some predict significant workforce reductions due to AI, experts like Sebastien de Brouwer from the European Banking Federation suggest that labor shortages, aging populations, and strong union representation in Europe may lead to retraining rather than mass layoffs. Productivity statistics have not yet shown the dramatic boom that would indicate widespread AI-driven job replacement. Additionally, AI presents new risks, as evidenced by a record 444,000 fraud reports last year, with criminals increasingly using AI to exploit vulnerabilities in banking and online accounts, compelling banks to embed real-time AI solutions for protection.
The UK Financial Conduct Authority (FCA), through the Mills Review, is grappling with how to regulate AI in retail financial services. There's a debate about whether existing regulations are suitable for "agentic AI," which could provide valuable help to millions of investors. Some argue that overly strict regulation could prevent consumers from accessing helpful AI tools for information on investments, savings, and other financial areas. The FCA is urged to consider more radical changes and work with providers to establish guiding principles, rather than prematurely regulating AI as equivalent to traditional regulated activities.
