Yesterday, 26 May 2016, I had the honour of taking part in the Fifth Italian-Norwegian Conference on the “Impact of EU Regulation on the Banking and Financial Sector” – at the invitation of the Italian Norwegian Research Institute for Law and Economics (INRILE)– with a presentation on the “Impact of Artificial Intelligence on the Financial Sector”. As is customary on this blog, I am offering readers a summary of that presentation.
Given the significance, complexity and novelty of the impact of artificial intelligence on the financial sector, it is advisable to proceed with caution and to frame the issue around three concentric circles, moving from the general to the specific:
First: The general context on Artificial intelligence with reference to its concept and regulation
Second: The special context on regulation with reference to the European Artificial Intelligence Act of 2024
Third: Specific aspects of the impact of artificial intelligence on the financial sector
A) General context: Artificial intelligence. Concept and regulation
(a) The concept of an artificial intelligence (AI) system as a prodigious alembic
It is appropriate to begin this speech by explaining the concept of an artificial intelligence (AI) system, using the metaphor of the alembic to provide an extremely simple image of an extremely complex regulation. To do so, we must start with the definition of an AI system found in Article 3.1) of the European Artificial Intelligence Act (EAIA) as ‘a machine-based system designed to operate with varying levels of autonomy, which may show adaptability after deployment and which, for explicit or implicit purposes, infers from the input information it receives how to generate output information, such as predictions, content, recommendations or decisions, which may influence physical or virtual environments’.
Given that this definition refers to machines and suggests liquid and gaseous physical spaces, we move towards the image of the alembic because we believe it can offer a useful metaphor for explaining and understanding the information management method underlying the phenomenon of AI and the systems that generate it. If we refine our reasoning further, we realise that we are in the world of information regulation, and, therefore, nothing seems more certain than to turn to the definition of alembic offered by the Dictionary of the Spanish Royal Academy of Language as ‘a tool used to distil a volatile substance, consisting mainly of a container for heating the liquid and a conduit through which the distilled substance exits’.
If we take our reasoning one step further, we can conclude by defining AI systems ideographically as ‘alembic distillers that process large amounts of data or raw information stored in the input channel to generate—through an algorithmic transformation method that allows them to interact with their environment and learn accordingly—an output product in the form of net or distilled information.’
(b) AI pathology: “AIpendency” and “AIddiction”
In recent times, social psychology in developed societies has revealed a new pathological syndrome that manifests itself as a mild condition known as “AIpendency” (“a compulsive need for AI in order to experience its effects or alleviate the discomfort caused by its deprivation”) or a severe condition known as ‘AIddiction’ (‘dependence on AI that is harmful to health or mental balance’). This mainly affects young people who place expectations on AI applications—such as ChatGPT—that they are not designed for and are not intended to fulfil, such as serving as medical, nutritional or emotional confidants, and which can lead to their alienation due to their own ability to manipulate feelings.
(c) Should AI be regulated?
Once AI has been defined, the question arises: Should AI be regulated? Our answer must be yes, because the regulation of AI and its uses is necessary to ensure that the obvious advantages this technology is already bringing to humanity are not overshadowed by the abuses that are also occurring. Indeed, the omnipresence of AI in our global society is as useful as it is dangerous when its use degenerates into abuse in the form of manipulation of citizens’ behaviour through AI systems and models that are contrary to the basic human rights of thought and expression.
(d) How should AI be regulated?
The next question we must ask and answer is: How should AI be regulated? On the global stage, we find two basic models of AI regulation that can coexist. Indeed, the plasticity of AI itself requires that ‘hard law’ rules be supplemented by ‘soft law’ regulations which, in the form of voluntary codes of conduct, specify the criteria for the proper use of AI among homogeneous population groups (e.g., professionals in the same trade, such as doctors or lawyers). One example is the ‘Report and Recommendations of the Working Group on Artificial Intelligence’ approved on 6 April 2024 by the House of Delegates of the New York State Bar Association (NYSBA).
(e) The four models of AI regulation
In the international context, four models of AI regulation can be distinguished: a) The European model of unified mandatory regulation based on the European Artificial Intelligence Act (EAIA) (Regulation (EU) 2024/1689), which establishes harmonised rules on artificial intelligence. b) The Chinese model of state control based, since 2022, on three types of forceful intervention mechanisms: the regulation of algorithms, rules against -alleged- hoaxes and a sophisticated system of public authorisations. c) The model of some G7 members, based on voluntary adherence to certain codes of conduct. d) The US model of scattered regulation through state laws, presidential executive orders, federal agency guidelines, and chip export controls. The Executive Order of 11 December 2025 by the President of the United States on Artificial Intelligence impacts this latter model in order to deregulate it, liberalise it and prevent the fragmentation of state laws.
B) Special context: The European Artificial Intelligence Act
The European Artificial Intelligence Act (EAIA) is incorporated into Regulation (EU) 2024/1689 of the European Parliament and of the Council of 13 June 2024 laying down harmonised rules on artificial intelligence.
(a) Objective
The objective of the EAIA is twofold, as it consists of: Stimulating the benefits of AI by improving the functioning of the internal market, promoting the adoption of human-centred and trustworthy artificial intelligence (AI); and preventing the risks that the use of AI systems presents in the form of harmful effects in the EU.
(b) Scope of application
The scope of application of the EAIA is established in Article 2 through the following three operations:
(b.1) It establishes the inclusion of subjects and systems in two ways: It establishes full subjective inclusion covering operators of AI systems who are suppliers placing AI systems on the market or putting them into service, or placing general-purpose AI models on the market in the EU; those responsible for the deployment of AI systems that are established or located in the EU; importers and distributors of AI systems; manufacturers of products that place an AI system on the market or put it into service together with their product and under their own name or brand; authorised representatives of suppliers not established in the EU; and affected persons located in the EU. It also provides for the partial objective inclusion of certain high-risk AI systems because they will be subject to the LEIA with limited scope.
(b.2) It establishes exclusions based on two types of classification criteria for AI systems, such that: It establishes functional exclusions, because the LEIA will not apply to AI systems used for the following purposes: military, defence or national security purposes; general scientific research and development purposes; research, testing or development purposes relating to the AI systems or AI models themselves. Establishes subjective exclusions, because the LEIA will not apply to the following categories of subjects: public authorities of third countries or international organisations falling within its scope, intermediary service providers and individuals.
(b.3) Adds compatibility provisions because the application of the LEIA will be compatible with Union law on the protection of personal data, privacy and the confidentiality of communications; with the rules laid down by other Union legal acts relating to consumer protection and product safety; and with the legal, regulatory or administrative provisions of both the EU and its Member States that are more favourable to workers.
Note: Readers interested in EAIA can consult our book on ‘The European Law on Artificial Intelligence’ (“La Ley Europea de Inteligencia Artificial. El Reglamento (UE) 2024/1689 de 13 de junio de 2024”) published 12.05.2026 by American Publishers (ISBN-10 1972989502, ISBN-13 978-1972989500) available on Amazon as a Kindle edition and paperback.
C) Specific aspects of the impact of artificial intelligence on the financial sector
The European Parliament’s Resolution of 25 November 2025 ‘on the impact of artificial intelligence on the financial sector’ was published in the Official Journal of the European Union on 24 April 2026 (Ref. P10_TA(2025)0286, 0282025/2056(INI)) C/2026/1701).
If we focus our attention on the European Parliament’s recommendations for ensuring the responsible use of AI in financial services, we can highlight the following two points:
(a) Earlier warning from the European Parliament regarding the EU’s competitive disadvantage vis-à-vis the US due to insufficient investment in AI
The EPR begins these Recommendations (paragraphs 17 to 25) by:
(a.1) Lamenting the EU’s competitive disadvantage vis-à-vis the US, which means that the EU “is falling behind in terms of innovation and investment in AI, as demonstrated by the €33 billion in venture capital received by EU companies developing foundational models between 2018 and 2023, compared to the more than €120 billion received by their US counterparts;
(a.2) Considerimg that the financial services sector, as the sector with the highest expenditure on ICT services and products, has the potential to act as a catalyst in mobilising private investment in AI;
(a.3) Yrges, in this context of sluggish investment in AI within the Union’s financial sector, the presentation of an ambitious proposal to revitalise the European venture capital ecosystem, within the framework of the Savings and Investment Union”.
(b) Requests from the European Parliament to the Europen Union Institutions Commission, the Member States and the European and national financial market supervisory authorities
The EPR continue to submit requests to the European Union Institutions as the Commission, the Member States and the European and national financial market supervisory authorities so that they adopt the following ten measures (in the form of a decalogue):
(b.1) “to remove barriers to entry into the Union for innovative AI-driven financial firms, including by streamlining their licensing processes,
(b.2) facilitating their cross-border expansion and ensuring their inclusion in innovation hubs within the supervisory framework”.
(b.3) “to provide clear and practical guidance, drawn up in consultation with European and national supervisory authorities and stakeholders, on the application of existing financial services legislation in relation to the use of AI”
(b.4) “to examine how AI-driven tools can be used in financial markets, such as in brokerage, portfolio management and compliance automation, to contribute to the objectives of the Savings and Investment Union, in particular by supporting retail investors in making informed investment decisions”
(b.5) improving financial education, fostering innovation among firms, reducing market fragmentation and ensuring a safe environment for consumers” (…).
(b.6) “to support the responsible adoption of AI by promoting consistent interpretations and a proportionate application of current regulations; it considers that appropriate regulation of the deployment of AI in the financial services sector fosters the adoption of AI and public confidence in it (…)”.
(b.7) “to assess the added value of specific AI-focused controlled testing environments, innovation hubs and cross-border testing environments for financial services in enabling experimentation with AI-driven financial innovation, both to help start-ups test their products and to allow established institutions to explore new uses in a controlled environment, whilst safeguarding consumer protection and market integrity;
(b.8) “considers that making proper use of controlled AI testing environments could provide the structured and supervised testing environment needed to facilitate innovation and the responsible deployment of AI in the financial services sector”
(b.9) encourages European and national supervisory authorities to enhance supervisory tools and technology (SupTech) through the use of AI and to integrate them into day-to-day supervisory activities to improve the efficiency and effectiveness of financial supervision;
(b.10) notes that these tools are intended to support, rather than replace, human supervisors
Note: Readers interested can consult our blog post (ajtapia.com) dated 8 May 2026 on the “Impact of artificial intelligence on the financial sector: European Parliament resolution of 25 November 2025 (OJEU 24.4.2026)”