OECD - Artificial intelligence and open finance

OECD Artificial Intelligence Papers Artificial intelligence and open finance Synergies, trade-offs and policy implications No. 612 

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Disclaimer This work is issued under the responsibility of the Secretary-General of the OECD and does not necessarily reflect the official views of OECD Member countries. This document, as well as any data and map included herein, are without prejudice to the status of or sovereignty over any territory, to the delimitation of international frontiers and boundaries and to the name of any territory, city or area.

Photo credits: © Getty Images/greenbutterfly.

©

OECD 2026.

Attribution 4.0 International (CC BY 4.0). This work is made available under the Creative Commons Attribution 4 .0 International licence. By using this work, you accept to be bound by the terms of this licence (https://creativecommons.org/licenses/by/4.0/). Attribution – you must cite the work. Translations – you must cite the original work, identify changes to the original and add the following text: In the event of any discrepancy between the original work and the translation, only the text of original work should be considered valid. Adaptations – you must cite the original work and add the following text: This is an adaptation of an original work by the OECD. The opinions expressed and arguments employed in this adaptation should not be reported as representing the official views of the OECD or of its Member countries. Third-party material – the licence does not apply to third-party material in the work. If using such material, you are responsible for obtaining permission from the third party and for any claims of infringement. You must not use the OECD logo, visual identity or cover image without express permission or suggest the OECD endorses your use of the work. Any dispute arising under this licence shall be settled by arbitration in accordance with the Permanent Court of Arbitration (PCA) Arbitratio n Rules 2012. The seat of arbitration shall be Paris (France). The number of arbitrators shall be one. 3

sound governance and clear accountability, proportionate pre-contractual due diligence an d risk assessment, and robust contractual safeguards (including rights relating to access , audit, data protection and termination) supported by continuous monitoring over the life of third-party arrangements. 3 The term “Agentic” in this context refers to these models’ agency, or their c apacity to act independently and purposefully. 4 Legal Entity Identifier (LEI) enables clear and unique identification of legal entities participating in financial transactions and other official interactions (see https://www.gleif.org/en/organizational-identity/introducingthe-legal-entity-identifier-lei).OECD Artificial Intelligence Papers Artificial intelligence and open finance: Synergies, trade-offs and policy implications No. 61 As finance is being reshaped by a range of technological, regulatory and market developments, AI and open datasharing are two particularly influential trends. However, dynamics of their intersection remain relatively understudied. This paper examines the interplay of AI innovation with data-sharing environments, highlighting mutually reinforcing benefits alongside increased complexity, trade-offs and amplified risks. It also explores a forward-looking theoretical scenario of agentic AI in an environment of growing data-sharing. The paper aims to support the responsible and scalable deployment of AI innovation within open finance ecosystems.

Any dispute arising under this licence shall be settled by arbitration in accordance with the Permanent Court of Arbitration (PCA) Arbitratio n Rules 2012. The seat of arbitration shall be Paris (France). The number of arbitrators shall be one. 3

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Abstract As finance is being reshaped by a range of technological, regulatory and market developments, Artificial Intelligence (AI) and open data-sharing are two particularly influential trends. However, the dynamics of their intersection remain relatively understudied. This paper examines the interplay between AI innovation and data-sharing environments, highlighting mutually reinforcing benefits alongside increased complexity, trade-offs and amplified risks. It also explores a forward-looking theoretical scenario of agentic AI in an environment of growing data-sharing. The paper aims to support the responsible and scalable deployment of AI innovation within Open Finance ecosystems.

JEL codes: G53 Financial Literacy; G510 Household Finance: Household Saving, Borrowing, Debt , and Wealth; O33 Technological Change: Choices and Consequences; Diffusion Processes.

This paper is part of the series “OECD Artificial Intelligence Papers”, https://doi.org/10.1787/dee339a8-en4 

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Acknowledgements This report provides analysis of the interplay and mutual dependencies between the deployment of AI in the financial sector and the evolving data-sharing frameworks. The report bui lds on prior OECD analysis on AI in finance and Open Finance as standalone digitalisation trends (OECD, 2023 [1]; 2023 [2]; 2021 [3]; 2023[4]; 2024 [5]; 2026 [6]) (2026[7]). The analysis discusses how these two trends mutually reinforce each other’s development, generating new opportunities while simultaneously giving r ise to new complexities and potential trade-offs, and amplifying certain risks. The paper also explores benefits and risks associated with a scenario of the proliferation of Agentic AI in finance, given the relevance of pers onal information

other’s development, generating new opportunities while simultaneously giving r ise to new complexities and potential trade-offs, and amplifying certain risks. The paper also explores benefits and risks associated with a scenario of the proliferation of Agentic AI in finance, given the relevance of pers onal information enabled through AI for such highly autonomous systems. The objective of this report i s to examine interdependencies of Open Finance and AI and associated benefits, challenges and trade-offs, with a view to fostering wider deployment and scale-up of responsible and safe AI innovation within Open Finance-enabled ecosystems. The report has been developed by the Capital Markets and Financial Institutions of the OECD Directorate for Financial and Enterprise Affairs. It was drafted by Iota Kaousar Nassr under the supervisi on of Fatos Koc, Head of the Financial Markets Unit, and Serdar Çelik, Head of Division. Liv Gudmundson and Mathilde Le Pichon provided editorial and communication support. The authors gratefully acknowledge valuable input and feedback provided by the following individuals and organisations: Merel Croon, Dutch National Bank; Gerardo García, Banco de México and Homero Issac Cardenas Escalante, Undersecretary of Finance and public credit, Mexico; Paweł Gąsiorowski and Adam Głogowski, National Bank of Poland; Mládek Josef Ing., Ministry of Finance of the Czech Republic; Mikari Kashima, Bank of Japan; Marina Kalfić, Ministry of Finance, Slovenia; Nishad Majmudar and Paull Randt, U.S. Department of the Treasury; Alexis Noir-Luhalwe, Direction Générale du Trésor, Franc e;Jungphil Park, Bank of Korea; Andrea Quiroga Angel and Maria Paula Rueda Viviescas, Financial Superintendency of Colombia; Natalia Radichevskaia, Luxembourg; Ryosuke Ushida and Kodama Kunihiro, Fi nancial Services Agency, Japan; Silvia Vori and Giuseppe Gra nde, Banca d’Italia; as well as Luis Aranda,

of Colombia; Natalia Radichevskaia, Luxembourg; Ryosuke Ushida and Kodama Kunihiro, Fi nancial Services Agency, Japan; Silvia Vori and Giuseppe Gra nde, Banca d’Italia; as well as Luis Aranda, Giuseppe Bianco, Maria Canedo, Celine Caira, Clarisse Girot, Miles Larbey, Beatriz Ma rques, Richard May and Nikolas Schmidt from the OECD. The report was discussed by the OECD Committee on Financial Markets, chaired by Mr Seiichi Shimizu, Assistant Governor, Bank of Japan, on 11 September 2025 and 5 March 2026. The report constitutes part of the horizontal OECD project on Artificial Intelligence. 5

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Table of contents Disclaimer 2

Abstract 3 Acknowledgements 4 Executive summary 7 1 AI and Open Finance: Complementarities and synergies 10 1.1. AI and Open Finance complementarities and synergies: Data as the common thread 10 1.2. Open Finance as an enabler for AI in finance: Inclusion and hyper-personalisation 11 1.3. AI innovation accelerating Open Finance objectives, under certain conditions 15 2 Risks and unintended consequences 20 2.1. A multifaceted risk landscape 20 2.2. Risk of overreliance by retail financial consumers and other consumer-related risks 21 2.3. Amplification of data governance concerns 22 3 Potential trade-offs arising from the interplay of AI and Open Finance 25 3.1. Model performance and data minimisation 25 3.2. Safeguards to local data and fragmentation 26 3.3. Trade-offs associated with the involvement of BigTech in Open Finance 27 4 Data sharing in a theoretical scenario of Agentic AI proliferation 29 4.1. Agentic AI experimentation in finance 29 4.2. The upside: Proactive hyper-personalised financial advice enabled by Open Finance and Agentic AI 29

4 Data sharing in a theoretical scenario of Agentic AI proliferation 29 4.1. Agentic AI experimentation in finance 29 4.2. The upside: Proactive hyper-personalised financial advice enabled by Open Finance and Agentic AI 29 4.3. The downside: Lack of alignment, accountability, liability and potential loss of human control 326 

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

5 Preliminary policy considerations 33 References 35 Annex A. Update to Open Finance frameworks since 2023 OECD report 40 Notes 42

FIGURES Figure 1. AI and Open Finance: synergies and mutually reinforcing benefits 11 Figure 2. Interplay between AI and Open Finance: Amplified risks 20 Figure 3. Interplay between AI and Open Finance: Trade-offs and tensions 25 Figure 4. Interplay between AI and Open Finance in a hypothetical scenario of Agentic AI proliferation in Finance 31

TABLES Table 1. Traditional versus “AI-ready” Data Governance 14

BOXES Box 1. Open Finance frameworks 12 Box 2. “AI Ready” Data governance in the AI era: Bank of Korea case study 14 Box 3. Open Finance in Brazil: Framework, use cases and success factors 17 Box 4. Data privacy considerations in Open Finance frameworks 19 Box 5. Risks of unfair consumer outcomes in data-sharing frameworks and the role of the G20/OECD HighLevel Principles on Financial Consumer Protection 22 7

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Executive summary The interplay between artificial intelligence and Open Finance The global finance sector is undergoing a period of significant digital tran sformation, catalysed by the increasing deployment of artificial intelligence (AI) and the evolving framew orks of data-sharing, such as Open Finance arrangements.1 The proliferation of AI in financial services has far-reaching implications for

The global finance sector is undergoing a period of significant digital tran sformation, catalysed by the increasing deployment of artificial intelligence (AI) and the evolving framew orks of data-sharing, such as Open Finance arrangements.1 The proliferation of AI in financial services has far-reaching implications for financial markets and their participants, while the evolution of data-sharing framewor ks reshapes the way consumers and financial firms share financial data. The individual impl ications of AI and Open Finance have been the subject of extensive analysis ; however, the intersection of these two transformative trends remains under-explored. This analytical gap warrants closer examination, given the potential for their interplay to generate novel dynamics for financial intermediation, with implications for policymakers. A mutually reinforcing beneficial relationship Open Finance provides the foundational infrastructure and critical data flows necessary to enable greater interoperability across the financial sector, serving as a key enabler for the effec tive deployment of AI in finance. It creates the conditions under which AI systems can operate optimally , by enhancing model capabilities to deliver hyper-personalised outputs tailored to the specific profiles and needs of individual financial consumers. AI model development and training, powered by the broad data access enabled under data-sharing frameworks, can unlock insights and value, driving competitive advantage through efficiency and productivity gains, hyper-personalisation and customisation of financial services, and broader financial inclusion. This relationship is mutually reinforcing, with synergies observed on both sides. While Open Finance empowers AI through access to high-quality structured data, AI in turn unlocks the analytical potential of Open Finance data flows , supporting the achievement of core O pen Finance objectives (financial inclusion, innovation and competition, empowerment of consumers through control of their data). Risk amplification at the intersection of AI with Open Finance The intersection of these two trends is accompanied by the amplification of challenges and risks found in each of the two trends when considered independently . Risks related to bias and discrimination may be exacerbated when data is used beyond the parameters of informed customer consent. Transparenc y challenges may become more pronounced, as individuals may be unable to track the downstream use of

each of the two trends when considered independently . Risks related to bias and discrimination may be exacerbated when data is used beyond the parameters of informed customer consent. Transparenc y challenges may become more pronounced, as individuals may be unable to track the downstream use of their data in subsequent AI model iterations. In other words, how their data, shared through Open Finance arrangements for some purposes, could be subsequently used in iterative AI model iterations, part icularly given the limited explainability of advanced systems. Such opacity may further complicate risk management in areas including data governance and data protection. Reduced user control over data could emerge as a result of (and depending on) the scale and integration of data embedded within Agentic AI models.8 

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Navigating associated trade-offs and tensions The intersection of AI within Open Finance frameworks also involves interesting tr ade-offs and potential tensions. These include potential tensions between model performance and data minimisation objectives; consent management and the iterative nature of model development; as well as the ne ed for data to be locally stored for AI training and possible fragmentation risks at the operat ional level (in other words, tension between operational costs and latency). Possible trade-offs could also be considered around the involvement in Open Finance frameworks of technology firms (BigTech) that are also operati ng in highly concentrated markets for AI models and infrastructure associated with AI model deployment. While their inclusion could further amplify concerns around third-party dependencies, their ou tright exclusion from data-sharing frameworks could significantly dampen incentives for innovation by narrowing the set of actors that can leverage rich financial and cross-sector data to create new products and services. Exploring a scenario of Agentic AI proliferation in data-sharing environments AI Agents are systems that can perceive and act upon their environment with a degree of autonomy, using tools as needed to achieve specific goals and adapt to changing inputs and contexts. Agentic AI generally refers to systems composed of multiple co-ordinated AI agents that can break down tasks, collaborate and

AI Agents are systems that can perceive and act upon their environment with a degree of autonomy, using tools as needed to achieve specific goals and adapt to changing inputs and contexts. Agentic AI generally refers to systems composed of multiple co-ordinated AI agents that can break down tasks, collaborate and pursue complex objectives autonomously over extended periods (OECD, 2026 [8]). Under a theoretical scenario, Agentic AI systems powered by Open Finance-enabled data could redefine how individuals and businesses interact with financial services, enabling a shift towards more proactive, automated, 24/7 hyper-personalised financial management, with reduced reliance on direct human intervention. Firms that can access data through data-sharing frameworks could unlock significant valu e for their customers and gain a competitive advantage, while also benefiting end customers through improved quality, diversity, and terms of financial products and services. A scenario of Agentic AI proliferation could involve important potential benefits for retai l investor participation in more complex capital market products, through improved access, potentially reduced costs linked to increased efficiencies and automation, real-time advisory capabilities, and personalised decisionsupport tools. At the same time, such a scenario also involves possible risks of overreliance on algorithmic outputs, potentially reduced human control over financial decisions and the risk of misaligned incentives, with ethical (and broader societal) implications. It should be noted that this remains a theoretical scenario. Unlocking synergies and mitigating risks through sound governance The transformative potential of AI in an era of increasing data-sharing through O pen Finance-type arrangements can only safely materialise if associated risks are identified and adequately mitigated . Effectively harnessing their synergies in a responsible manner will be key to achieving the multiple benefits at the intersection of Open Finance and AI. It will depend on the existence of robust governance structures that emphasise human oversight and accountability , and on compliance with appropriate rules and standards. Safeguarding data security, privacy and obtaining explicit consumer consent remain important, especially given the critical role that such data plays in the development and deployment of AI systems. A proportionate, risk-based approach to regulatory implementation can help prevent unintended s tifling of

standards. Safeguarding data security, privacy and obtaining explicit consumer consent remain important, especially given the critical role that such data plays in the development and deployment of AI systems. A proportionate, risk-based approach to regulatory implementation can help prevent unintended s tifling of innovation caused by overly complex regimes (including privacy frameworks, among others). Cross-sector collaboration, particularly between AI, the financial sector and privacy policy communities, is vital to address challenges in a coherent manner, supported by OECD instruments that provide guiding principles for data governance and privacy . Enhancing consumer awareness through financial education and targeted outreach campaigns can empower individuals to exercise and manage their data rights effectively within AI-enabled Open Finance environments. Going forward, it would be ben eficial to conduct a fact- 9

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

finding exercise to identify opportunities and challenges of existing data-shari ng systems in the era of AI development. Countries could benefit from the sharing of good practices on how fi nancial firms can build coherent governance frameworks that address issues combining data security and personal data protection across both Open Finance and AI domains and on how financial authorities can balance the promotion of AI innovation on the basis of data sharing frameworks while ensuring security and trust.10 

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

1.1. AI and Open Finance complementarities and synergies: Data as the common thread Although there is no common definition for data-sharing frameworks, Open F inance is commonly used to refer to the sharing, access, and reuse of customer financial data, both personal and non-personal, across a wide array of financial products and providers, through secure digital channels and with customer consent (OECD, 2023 [9]). Open Finance is an evolution of the Open Banking concept, expanding data-s haring beyond payment accounts into the full range of financial services, includ ing sectors like insurance, pensions, investment, and securities (see Box 2.2). The types of data included in such expanded scope of

provides the connective tissue enabling data flows and interoperability across the financial sector. By facilitating secure and standardised data-sharing, Open Finance can create the conditions under which AI model development and fine-tuning can be further enhanced and optimised, particul arly for personalised output and related products and services. In turn, AI offers advanced analytical capabilities to maximise the benefit from the use of data made accessible through Open Finance frameworks, while also advancing the broader goals of Open Finance frameworks related to more open, efficient, competitive and inclusive financial systems. The availability of high-quality data can also be considered a source of competitiveness for generated models (OECD, 2024[13]). For example, data used to train foundation models can be assessed for its impact on model competitiveness. Beyond minimum data requirements, factors such as data qu ality, scale/volume, uniqueness, variety and value are some of the dimensions that can imp act the 1 AI and Open Finance: Complementarities and synergies 11

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

competitiveness of generated or trained models. What is more, the use of large Ope n Finance datasets may also enhance the ability of participating firms to produce competitive open-weight models. Figure 1. AI and Open Finance: synergies and mutually reinforcing benefits

1.2. Open Finance as an enabler for AI in finance: Inclusion and hyperpersonalisation As Open Finance includes the sharing, access and reuse of personal and nonpersonal data (Box 1), it could act as a critical enabler for AI model development, serving as a catalyst for the availability of richer, more diverse, and interoperable datasets. Furthermore, such datasets could be used for AI model training and fine-tuning. By enabling access to a broader and more granular set of financ ial data, Open Finance enhances the informational foundation upon which AI models operate to deliver outputs pertinent to specific end-users and contexts . In particular, Open Finance can help address potential data gaps in financial analysis performed by AI-driven models, enhancing the reliability and accuracy of decision making

(OECD, 2023 [9]). Open Finance is an evolution of the Open Banking concept, expanding data-s haring beyond payment accounts into the full range of financial services, includ ing sectors like insurance, pensions, investment, and securities (see Box 2.2). The types of data included in such expanded scope of data sharing include data related to savings, mortgages and other consumer credit products, pensions and retirement funds, insurance policies (life, non-life, health), and investments ( stocks, bonds, funds). This broader scope of Open Finance compared to Open Banking frameworks expands data-sharing beyond payments data, aiming to encompass a financial consumer’s entire financial footprint. Artificial Intelligence (AI) systems are being adopted across all areas of financial market activity, including banking, insurance, asset management and securities markets to analyse data, automate task s, and enhance decision making processes (OECD-FSB, 2024 [10]). “Narrow” AI models based on Machine Learning (ML) have been used in finance for decades, while more advanced Generative AI (GenAI) models, such as Large Language Models (LLMs) are being increasingly deployed in finance (OECD, 2021[11]; 2023[12]). AI innovation enables financial institutions to improve efficiency, accuracy , and speed across a wide range of operations, while also improving customer interaction and support, as well as the development and delivery of financial services and products. Initial implementati ons of AI predominantly focussed on the automation of repetitive tasks, data analytics and predictive ins ights to support decision making, aiming to increase efficiency and reduce costs. Advanced AI-driven systems support increasingly complex and real-time analytics with evolving degrees of autonomy. The common thread underpinning both Open Finance and AI digitalisation trends is the central role of data, which lies at the core of both innovations. Data is the bloodline of AI innovati on, while Open Finance provides the connective tissue enabling data flows and interoperability across the financial sector. By facilitating secure and standardised data-sharing, Open Finance can create the conditions under which AI model development and fine-tuning can be further enhanced and optimised, particul arly for personalised

enhances the informational foundation upon which AI models operate to deliver outputs pertinent to specific end-users and contexts . In particular, Open Finance can help address potential data gaps in financial analysis performed by AI-driven models, enhancing the reliability and accuracy of decision making by substituting for missing or incomplete inputs, also supporting broader financi al inclusion objectives. In theory, the availability of more representative datasets of clients can also help tra in or fine-tune more accurate and nuanced models, possibly allowing for some reduction in biases that might arise from limited, fragmented or skewed datasets. OF interoperability and utility AI automation supporting data normalisation/ standardisation 02

CAVEAT: materialised only if safeguards are in place

AI Accelerator for Open Finance OPEN FINANCE AI enabler Data analysis AI unlocking analytical potential of OF dataflows Support delivery of OF objectives Promote financial inclusion (e.g. credit scoring) Stimulate innovation and competition Foster consumer empowerment through control of own data AI model training and fine-tuning Richer, more diverse, and interoperable datasets AI model output More accurate and nuanced Hyper-personalisation Tailored financial product/service provision Hyper-personalised outputs (e.g. robo-advice for investment) Data at the core12 

ARTIFICIAL INTELLIGENCE AND OPEN FINANCE © OECD 2026

Although synthetic data is a powerful supplement to real data, especially when real data is scarce or expensive to label , empirical evidence suggests that the use of real data results in more robust model outputs when compared to synthetic datasets (Singh et al., 2024[14]). In particular, the synthetic clones are much more susceptible to adversarial and real-world noise than models trained with real data. Hybrid approaches, leveraging both synthetic and real data, tend to yield the most robust models. The use of AI for creditworthiness assessment is a prime example of such a positive influence, particularly beneficial for thin-file borrowers such as small and medium-sized enterprises (SMEs) without prior formal

approaches, leveraging both synthetic and real data, tend to yield the most robust models. The use of AI for creditworthiness assessment is a prime example of such a positive influence, particularly beneficial for thin-file borrowers such as small and medium-sized enterprises (SMEs) without prior formal credit history or collateral. AI-driven credit scoring can use data available through data-sharing frameworks to generate holistic customer insights that can help lenders assess such prospective borrowers, potentially Box 1. Open Finance frameworks Open Finance includes the sharing, access and reuse of personal and non-personal data for the purposes of providing a wide range of financial services. In addition to providing a s ecure and privacypreserving framework where customers can consent to third parties accessing their data, Open Finance, similar to Open Banking, allows third parties to initiate payments and transactions or take other related actions on the customers’ behalf. There is no real separation between Open Bankingand Open Finance-related frameworks, and they can co-exist depending on the use cases. The evolution of data-sharing may ultimately extend to customer-permissioned access to relevant non-financial data (e.g. utility p ayments, telecom usage, or e-commerce transactions) further enriching the data landscape and enabling more holistic financial services. Key objectives of Open Finance frameworks include: • Fostering Innovation: by facilitating access to more comprehensive and diverse datasets, Open Finance supports the development of novel financial products, services and business models, including more personalised financial advice and offerings. • Enhancing Market Competition: b y reducing data concentration among incumbent institutions, Open Finance enables new entrants to compete on more equitable terms and enables th e introduction of new firms with similar or novel business models. • Empowering Consumers: Open Finance strengthens individuals’ control over their financial data, allowing them to determine who can access their information and for what purposes. This, in turn, allows for enhanced service personalisation, comparability and user experience. • Advancing Financial Inclusion: by leveraging alternative data sources to assess creditworthiness or deliver financial advice, Open Finance holds the potential to extend services to underserved or excluded populations.

in turn, allows for enhanced service