Artificial Intelligence: Towards an Ambitious European Regulatory Strategy
Artificial intelligence (AI) is fundamentally transforming the healthcare and finance sectors, compelling European policymakers to establish a regulatory framework suited to the challenges of technological sovereignty. In a joint analysis, Gilles Babinet, entrepreneur and chairman of the Conseil national du numérique, and Pascale Seivy, financial regulation expert, argue for an ambitious European strategy built upon regulation rather than mere competition with American and Chinese giants.
The two experts highlight that generative AI is already disrupting banking and insurance professions, from fraud detection to the personalisation of wealth management advice. This technological acceleration comes as Europe struggles to produce champions capable of rivalling OpenAI, Google or Baidu. For Babinet and Seivy, the mistake would be to attempt to imitate these models: the European path must instead rely upon intelligent regulation that establishes trust and creates a differentiated competitive advantage.
This position sits within an already dense regulatory context. The European AI Act regulation, adopted in March 2024 and progressively entering into force since, establishes a risk classification and strict obligations for high-impact systems, notably in finance. The Financial Conduct Authority (FCA) and the Prudential Regulation Authority (PRA) have already published guidelines on the use of algorithms in asset management and credit. The Bank of England, for its part, is experimenting with AI for macroprudential surveillance and cyber threat detection.
In markets, this rising regulatory power creates characteristic tension. French and European financial institutions are investing heavily — more than €15 billion annually for the European banking sector according to 2023 estimates — yet remain confronted with growing compliance challenges. AI models, often described as "black boxes", struggle to satisfy the explainability requirements of the Markets in Financial Instruments Directive (MiFID II) and the forthcoming Digital Operational Resilience Act (DORA) on digital resilience. This friction between innovation and supervision could favour large systemic banks, better equipped to deploy legal and technical resources, to the detriment of fintechs and medium-sized asset managers.
The European Central Bank (ECB) main refinancing rate, maintained at 4 per cent despite initial cuts begun in June 2024, constitutes another structuring factor. In an environment of elevated capital costs, AI investments must demonstrate tangible return on investment. Financial institutions are thus redirecting expenditure towards high operational impact applications — back-office automation, enhanced credit scoring — rather than towards fundamental research in large language models (LLM).
For investors, this configuration offers two lines of reflection. On one hand, European technology stocks specialised in regulatory compliance (regtech) and algorithm explainability could benefit from a favourable structural tailwind, following the sector's performance since the adoption of GDPR in 2018. On the other hand, the emergence of a European "trustworthy" AI standard could constitute an export asset towards emerging markets seeking reference frameworks, notably in Africa and South-East Asia.
Babinet and Seivy's proposal nevertheless calls for caution regarding implementation timelines. The heterogeneity of national regulators within the Union, combined with political uncertainties linked to the 2024 electoral cycle, risks diluting the stated ambition. Investors will particularly monitor the first administrative sanctions handed down by the FCA and PRA regarding AI, which will establish the de facto jurisprudence of the new framework.
