What changed between the draft committee report and the plenary report

Removed:5. Notes the dependency of EU financial actors on third-country technology providers for software and AI tools, which may lead to concentration risk and reduce the bargaining power of financial institutions when negotiating or modifying contractual terms for AI services; emphasises that EU companies must be able to use existing cloud infrastructure for AI development and deployment; supports, however, EU initiatives to boost AI and cloud development;

Added:5. Highlights that the rise of AI poses challenges for supervisory authorities, particularly given the lack of AI-specific expertise and adequate supervisory tools to assess advanced machine learning and generative AI models; calls on the European and national supervisory authorities to adapt to the increasing use of AI in financial services and to monitor, assess and mitigate risks to consumers and financial stability, while being mindful not to discourage innovation through disproportionate compliance burdens or overly prescriptive regulatory approaches;

Added:6. Notes that the concentration among AI service providers that offer investment advice may lead to herd behaviour, driven by similar models and limited data sources; urges the European and national supervisors to monitor these risks and financial institutions in order to account for them when developing AI tools;

Added:7. Notes the dependency of EU financial actors on third party technology providers (TPPs) to host and develop their AI models and highlights that the majority of financial firms are reliant on only a few TPPs for these services, which may lead to concentration risk and reduce the bargaining power of financial institutions when negotiating or modifying contractual terms for AI services; cautions that reliance on a small number of providers for a given service could lead to systemic risks in the event of disruptions, especially if rapid migration to alternative providers is not feasible;

Added:8. Notes that the recently enacted Digital Operational Resilience Act (DORA) requires financial institutions to implement measures to mitigate concentration risk stemming from information and communication technology (ICT) TPPs, including contingency plans and arrangements to ensure business continuity; requests that the Commission and the European supervisory authorities assess, in particular, the feasibility of applying the exit strategies and transition provisions stipulated in DORA to AI models hosted by the infrastructure of TPPs, especially with regard to the considerable reliance on third-country TPPs for AI services;

Added:9. Emphasises that EU companies must be able to use existing cloud infrastructure for AI development and deployment; calls for actively exploring avenues to strengthen the compatibility and interoperability of AI models and compliance frameworks with those of like-minded international partners, especially those that aspire to provide equally robust regulatory safeguards, to ensure that EU financial institutions maintain access to AI tools and suppliers and with a view to shaping balanced global standards while safeguarding legal certainty for European businesses;

Added:10. Supports initiatives to boost AI and cloud development in the EU, especially with a view to developing AI services that are fully compliant with EU data protection and fundamental rights frameworks, while also strengthening strategic autonomy and resilience;

Changed:6.11. Stresses that the financial services sector is highly regulated, subject to multiple pieces of sectoral legislation at both national and EU level, requiring actors to manage risks in a variety of areas including data protection, data lineage, data quality, data governance, operational resilience, outsourcing, model risk, discriminatory outcomes, and market and credit risk, which together form the framework for AI deployment and governance in the financial services sector;sector16; emphasises, however, the importance of continuously monitoring regulatory gaps and evolving use cases of AI in finance, especially with a view to safeguarding consumer rights and the right to privacy;

Removed:7. Expresses concern about regulatory overlaps and legal uncertainties between the AI Act and sectoral legislation;

Added:12. Notes that the EU has adopted a more risk-based approach to AI regulation than other jurisdictions; underlines that, while this may create challenges for the adoption and development of AI in financial services, it also offers an opportunity to build trust and support innovation, provided that the framework is clarified and implemented in a way that fosters legal certainty, proportionality and market confidence; recognises that the AI Act has not yet been fully implemented and that its practical implications have not yet been assessed;

Changed:8.13. Recalls that the AI Act allowsexplicitly takes into account the current financial services acquis and seeks to avoid duplication of requirements, particularly with regard to internal governance and quality management processes, by allowing for limited derogations for financial institutions wherein so far as equivalent requirements existare underlaid down in EU financial services law; expresses concern aboutthat thethere are, nonetheless, regulatory overlaps and a lack of sufficient guidance on the interpretation of these regulatory overlaps and interactions;interactions, cautionswhich againstintroduces adoptingundue complexity, compliance burdens and legal uncertainty, thus hindering the uptake of AI in the financial services sector; underlines the importance of guaranteeing a maximalistlegal, approachregulatory toand administrative framework that is based on certainty, predictability and stability; notes that an expansive interpretation of the AI Act, whichrather wouldthan furthera complicateproportional one, may risk leading to undue compliance requirements for financial institutions;institutions and causing legal uncertainty; recognises the challenge arising from the fact that supervisory agencies have differing legal interpretations and expectations in terms of the application of the acquis, resulting in fragmentation of the single market; asks the Commission and the national competent authorities to identify and address any inconsistencies in the course of the AI Act’s implementation and as part of the upcoming Digital Omnibus package;

Removed:9. Expresses concern that the GDPR and its requirements on data minimisation, purpose limitation, customer consent, and financial institutions’ processing of personal data impose limitations to the use of AI in financial services;

Added:14. Supports the AI Act’s recommendation to designate financial competent authorities as market surveillance bodies for AI systems used in financial services deemed high-risk; notes, however, that other national competent authorities will be responsible for supervising non-high-risk AI systems; recognises the challenges arising from having multiple supervisory agencies with competences regarding the application of the acquis; recognises, furthermore, the challenges arising from the differing legal interpretations and expectations of the various supervisory agencies, which could lead to the fragmentation of the single market;

Added:15. Encourages the supervisory authorities to strengthen coordination, cooperation and information exchange to avoid overlapping jurisdiction claims; urges, moreover, the Commission and the supervisory authorities to strengthen cooperation with international partners in global standard-setting forums to ensure alignment and avoid the fragmentation of regulatory approaches, as well as to ensure that the EU keeps pace and aligns with global regulatory developments;

Added:16. Notes that the General Data Protection Regulation and its requirements on data minimisation, purpose limitation, customer consent, and financial institutions’ processing of personal data impose limitations on the use of AI in financial services; considers that the right balance is needed between reaping the benefits of the use of AI in financial services and the protection of consumers’ data;

Changed:10.17. Regrets that the EU is lagging behind in terms of AI innovation and investment;investment, as illustrated by the EUR 33 billion in venture funding received by EU companies developing foundational models between 2018 and 2023, compared to over EUR 120 billion received by their US counterparts17; believes that the financial services sector, as the largest spender on ICT services and products, has the potential to act as a catalyst in mobilising private investment in AI; calls, against the backdrop of slow AI investment in the EU’s financial sector, for an ambitious proposal to jump-start the European venture capital scene as part of the savings and investments union;

Changed:11.18. Calls on the Commission to ensureprovide clarityclear and guidancepractical guidance, developed in consultation with the European and national supervisory authorities and stakeholders, on howthe application of existing financial services regulationslegislation applywith regard to the use of AI in financial services;AI; considers that such guidance should aim to enable the use of AI in the financial services sector;sector, including in a way that is ethical, responsible and transparent; calls for consistent definitions and the simplification of the regulatory framework to avoid duplicated requirements, including risk assessment reporting requirements;requirements, and cautions against a one-size-fits-all approach that places a disproportionate burden on smaller and medium-sized financial institutions; emphasises the need for a good balance between the responsible use of AI and providing enough room for innovation;

Removed:12. Warns against the adoption of new sectoral legislation to regulate AI in financial services, as there are already established sectoral rules that cover AI deployment; believes that this would create additional layers of complexity and uncertainty and ultimately deprive the sector of the benefits of AI use; strongly advises the Commission and the Member States to coordinate to avoid gold-plating relevant legislation and to prevent the creation of new barriers in cross-border markets;

Added:19. Calls on the Commission to explore how AI-driven tools can be used in financial markets, such as in intermediation, portfolio management and compliance automation, to contribute to the objectives of the savings and investments union, including by supporting retail investors in making informed investment decisions, enhancing financial education, fostering innovation among companies, reducing market fragmentation and ensuring a safe environment for consumers; stresses that achieving these goals requires a technology-neutral regulatory framework;

Removed:13. Calls on the European and national supervisory authorities to support the uptake of AI by promoting consistent interpretations and avoiding overly strict application of existing regulations; believes that failure to support the uptake of AI risks depriving the financial services sector of the benefits of AI; emphasises that the attitude and approach of supervisors are as important as the rules themselves; recommends that supervisory efforts prioritise tangible, operational risks where identified, rather than abstract or theoretical concerns;

Added:20. Believes that sectoral legislation regulating the use of AI in financial services is mainly sufficient to cover AI deployment in its current form; underlines that there should be continuous monitoring to determine if there are duplications or deficiencies in the current financial services legislation applicable to AI deployment; underlines that additional legislation would add complexity and uncertainty and ultimately risk depriving the sector of the benefits of AI use; stresses that reliance on current frameworks requires continuous supervisory attention, effective enforcement and clear allocation of responsibility for ensuring compliance, particularly in cross-border or outsourced AI deployment scenarios, as well as the monitoring and assessment of possible future gaps created by new AI developments if they create substantial risks to consumers and financial stability; strongly advises the Commission and the Member States to coordinate to avoid gold-plating relevant legislation and to prevent the creation of new barriers in cross-border markets; notes that the Commission, according to the AI Act, can assess the list of high-risk applications under Annex III to the AI Act;

Removed:14. Believes that significant changes in the use of AI will require appropriate skills and talent, considering that the use of AI is dependent on human capabilities; supports industry measures to improve the understanding and responsible use of AI technology in the sector; asks for more clarity with regard to the AI Act’s requirements for financial institutions to comply with AI literacy requirements;

Added:21. Calls on the European and national supervisory authorities to support the responsible uptake of AI by promoting consistent interpretations and proportionate application of current regulations; believes that adequate regulation of AI deployment in the financial services sector supports uptake and societal trust in AI; emphasises that the attitude and approach of supervisors are as important as the rules themselves; recommends that supervisory efforts prioritise tangible, operational risks where identified, rather than abstract or theoretical concerns, while maintaining an active and proportionate approach to supervision, by balancing innovation and consumer protection, to manage unforeseen risks arising from the widening uptake of AI technologies; stresses the role of effectively monitoring and addressing AI-related risks, including those related to opacity, market concentration and loss of accountability, which could impact financial stability;

Added:22. Calls on the Commission and the Member States to remove entry barriers within the EU for AI-driven innovative financial undertakings, including through streamlined licensing, cross border scale-ups and inclusion in supervisory innovation hubs;

Added:23. Supports research into the environmental impact of AI use, with a focus on resource intensity and long-term sustainability, in order to increase transparency and help financial institutions to assess these aspects and their own environmental footprint;

Added:24. Believes that the increasing use of AI, which may have implications for the financial services job market, requires strong AI literacy, digital skills, and talent involvement, supported by both public-sector upskilling initiatives and market-based solutions; supports industry efforts and targeted initiatives, including public-private partnerships and reskilling programmes, to build technical and ethical AI competencies, especially regarding rights and risks, in the financial workforce; underlines the importance of developing AI strategies that enhance productivity, while supporting workers’ adaptation, upskilling and reallocation, while ensuring meaningful human oversight and control; asks for more clarity with regard to the AI Act’s requirements for financial institutions to comply with AI literacy requirements; stresses, furthermore, the importance of ensuring and promoting equal access to AI tools and services, including for less digitally capable segments of the population;

Added:25. Calls on the Commission and the European and national supervisory authorities to assess the added value of AI-specific regulatory sandboxes, 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 incumbent institutions to explore new use in a controlled setting, while safeguarding consumer protection and market integrity; believes that properly leveraging AI regulatory sandboxes could provide the structured, supervised testing environment necessary to facilitate innovation and responsible AI deployment within the financial services sector; encourages the European and national supervisory authorities to enhance supervisory tools and technology (SupTech) through the use of AI and integrate them into daily supervisory activities to improve the efficiency and effectiveness of financial supervision; notes that these tools are intended to support, not replace, human supervisors;

Changed:A. whereas the EU Artificial Intelligence Act (AI Act) introduces the world’s first comprehensive regulatory framework for artificial intelligence (AI);

Changed:B. whereas points 5(b) and (c) of Annex III to the AI Act definesdefine two high-risk use cases for the financial services sector, namely the use of AI systems for consumer credit scoring and creditworthiness assessments and their use for risk assessments and pricing of life and health insurance;

Changed:15.26. Instructs its President to forward this resolution to the CouncilCouncil, the Commission and the Commission.governments and parliaments of the Member States.

Changed:4. Understands that financial institutions have adopted a prudent and gradualmeasured approach to developingdeveloping, testing and deploying AI systems, with a view to ensuring compliance with existing cross-cutting and sectoral legislation; underlines that this prudent approach may also be driven by undemonstrated customer demand, evolving customer expectations and risk considerations;considerations, which have been outlined above;