In mid-June, IOSCO published the report "SupTech: Mapping the Use of Technology in Financial Supervision", setting out the findings of its inaugural SupTech Survey. Conducted during March and May 2025 under FINMA’s leadership, the survey covered 49 authorities across all IOSCO regions, representing more than three quarters of global securities market value, and examined the strategic, operational and institutional dimensions of SupTech adoption. The results indicate that SupTech is moving beyond experimentation and becoming integrated into core supervisory functions. Adoption is driven primarily by institutional needs — particularly efficiency, timelier information and stronger supervisory capabilities — while AI applications, improved data access and cloud infrastructure stand out as the main technology enablers. The survey also shows that the availability of technology alone is not sufficient: cyber and data security, third-party dependencies and operational risk remain prominent concerns, while limited funding constrains progress toward more advanced tools.

Survey responses also indicate that institutional arrangements are becoming more defined, although implementation remains uneven. Most authorities report that their SupTech, data and innovation strategies are still in a phase of partial implementation, even as dedicated funding and senior-level ownership become more common, particularly among smaller authorities and growth and emerging market regulators. Cooperation is broad and spans financial institutions and technology providers, foreign securities regulators, central banks, international organisations, government bodies and other partners. However, engagement with foreign regulators remains centred on exchanging experience and good practice rather than sharing code or technical tools, with legal, confidentiality and data protection constraints limiting deeper information sharing.

Current applications remain concentrated in consumer and investor protection and capital markets supervision, but interest is widening into less mature domains. Digital assets show the largest gap between current use and reported interest, indicating that authorities are exploring SupTech applications even where operational deployment remains limited. The underlying technology base remains pragmatic: most authorities rely on mid-level solutions, transaction data and established fraud detection tools, while aspiring to expand into advanced analytics and machine learning. Finally, responses point to human capital strategies remaining underdeveloped, with authorities relying mainly on online and ad hoc training, prioritizing external recruitment over internal capability building, and making limited use of structured mentoring.

Public announcements from Q1 and Q2 2026 indicate that the SupTech pipeline remains strong. While the pool of publicly disclosed initiatives remains highly diverse, several focus areas have crystallized. First, economic analysis and inflation forecasting remain among the more mature areas of AI experimentation in central banking. Recently disclosed work builds on an established base of machine learning, neural network, alternative data and text-based applications, but the emphasis is increasingly shifting towards operational use cases that can process larger datasets, support more frequent updates and incorporate less structured information. Project Spectrum, developed by the Bank for International Settlement's Innovation Hub, the Deutsche Bundesbank and the European Central Bank (ECB), reflects this direction by using AI-assisted classification to make high-frequency online price data more usable for inflation analysis. The European Central Bank also shared details on how its Corporate Telephone Survey, which involves the collection of qualitative information from large corporations about current economic developments, applies speech-to-text and large language models to streamline the processing and scoring of qualitative firm intelligence.

A second direction is the broadening of technology-enabled access to regulatory information and supervisory services. Qatar's Central Bank launched a website assistant for accessing published information, while the Hong Kong Mandatory Provident Fund Schemes Authority introduced a purpose-specific generative AI assistant grounded in its official investment education material. Globally, now 25+ authorities have public-facing virtual assistants in place to assist with general information queries, complaints handling or more specific support for licensing applicants. Alongside conversational tools, authorities are also making official content and external interactions more structured. Norway’s Financial Services Authority launched Tilsynsportalen, which uses AI to organise published supervisory reports for search, filtering and comparison. The new platform follows similar new capabilities introduced by other authorities during 2025 including natural language-powered search and information synthesis by the New Zealand Financial Markets Authority’s and machine learning assisted search by the Monetary Authority of Singapore.

Third, AI-based monitoring systems continue to gain traction. The Brazilian Pension Funds Authority (PREVIC) is testing an AI-based monitoring system, developed with the Federal University of Rio de Janeiro, to analyse the solvency and financial balance of supplementary pension plans. The platform uses PREVIC data to calculate plan-level risk indicators, summarise complex technical documents, assess assumptions used in liability valuations and project results over a three-year horizon. PREVIC expects the system to enter operation in September 2026. Separately, it is also developing an AI capability jointly with a start-up to analyse atypical investments, extending its use of advanced analytics across pension fund supervision. In Singapore, the Monetary Authority is conducting a proof-of-value with banking industry partners, the Government Technology Agency of Singapore and the Singapore Police Force to test AI and machine learning techniques for pre-emptive scam detection. The exercise brings together data from five banks to develop more robust models for identifying higher-risk transactions and accounts, with the aim of enabling earlier assessment, intervention and reduction of customer losses from scams.

Outside of core Suptech, authorities also continue to expand research efforts into the application of AI. A number of these involve testing AI in controlled settings to examine how models or AI-based agents respond to information, constraints and incentives and to derive policy, supervision and market implications. For example, the Bank for International Settlements (BIS), through its Innovation Hub London and Eurosystem Centres, and together with the Bank of England and the Deutsche Bundesbank, launched Project Logos to observe how large language model-based agents behave as portfolio managers in a simulated financial market environment, including how they interpret information, allocate capital and respond to constraints over time. Separately, a new working paper by the Spanish Securities Commission tested four LLMs in a ten month live stock selection exercise, finding that structured prompting, human oversight and grounding recommendations in official regulatory filings improved model performance, while recurring reasoning and dat -retrieval failures constrained reliable autonomous use.

In April 2026, Representatives Marlin Stutzman and Bill Foster introduced the bipartisan "Fostering the Use of Technology to Uphold Regulatory Effectiveness in Supervision Act (H.R. 8278)" which would establish a common assessment and reporting framework for supervisory technology across a defined group of US federal financial authorities. The bill seeks to address identified limitations in agencies’ access to real time information, continued reliance on outdated analytical tools and procurement constraints that can delay the modernisation of supervisory systems. It also points to the rapid expansion of AI use by financial institutions as increasing the need for agencies to have the technology, expertise and skills required to assess the resulting opportunities and risks. The bill would cover the Federal Reserve Board, Consumer Financial Protection Bureau, Federal Deposit Insurance Corporation, Department of the Treasury—including the Office of the Comptroller of the Currency and the Financial Crimes Enforcement Network—Federal Housing Finance Agency and National Credit Union Administration. The Securities and Exchange Commission and Commodity Futures Trading Commission are not included in its scope.

Within 180 days of enactment, each covered authority would be required to assess whether its core IT infrastructure, supervisory and market monitoring tools, and systems for data collection, storage, processing and security support adequate real-time supervision. Authorities would also need to review their procurement rules and identify opportunities to streamline the acquisition, development and testing of new systems. They would subsequently submit a coordinated report to Congress within 18 months of completing the assessments and every five years thereafter, covering their supervisory technology, procurement practices, technology workforce, information collection processes, data quality and interoperability, interagency information sharing and planned system upgrades. The House Financial Services Committee approved an amended version of the Act by a 52–0 vote on 13 May 2026 and it was subsequently reported to the House.