What's new
Overview
This deep dive examines the latest developments in artificial intelligence, including emerging adoption patterns, key risk watch points, evolving policy and supervisory guidance, and research on the economic impact of AI.
What's new
Polish Financial Supervision Commission issues recommendations on Frontier AI cyber risks for financial market entities
The Polish Financial Supervision Commission has issued recommendations telling financial market entities to reassess ICT risk management and operational resilience for cyber threats linked to Frontier AI. Firms should incorporate AI threat scenarios into DORA processes, prioritize vulnerability management, third-party risk and protection of internet-facing assets, and ensure senior management allocates sufficient resources. UKNF said the adequacy and effectiveness of firms' measures will be tested through ongoing supervision and inspections.
Payments Canada discusses agentic autonomous finance and the need for agent-aware payment rails
Payments Canada outlined a panel discussion on agentic autonomous finance and the implications of AI agents becoming active participants in payments and other financial activity. The update centers on identity, trust and control challenges, with discussion focused on tokenized assets, cryptographic trust and agent-aware payment rails. It also points to preparation needs for Canadian institutions and regulators.
Financial Conduct Authority selects 21 firms for second Supercharged Sandbox cohort, adds Anthropic support and launches Agentic Academy
The Financial Conduct Authority has selected 21 organisations for the second cohort of its Supercharged Sandbox and added Anthropic as a technology partner. Participants will use Claude tools to test advanced AI applications in areas including payments, fraud detection, governance, inclusion and compliance. The FCA also launched Agentic Academy, a 10-week AI programme for selected firms.
CDP launches optional AI-powered Suggested Response tool for the 2026 disclosure cycle, available to all disclosers from 23 July
CDP has introduced an optional AI-powered “Suggested Response” feature for its 2026 disclosure cycle, with rollout to all disclosers from 23 July 2026. The tool uses existing company documents to generate draft responses to CDP questionnaire items, while leaving organizations responsible for final review and submission. CDP said early testing showed 40% faster preparation and 25% higher response and completion rates.
Bank for International Settlements bulletin examines frontier AI cyber risks and urges swift defensive adoption
The Bank for International Settlements published a bulletin warning that frontier AI can strengthen both cyber offence and defence, but that cost asymmetries may leave attackers with an advantage. It says the impact on systemic cyber risk will depend on access to advanced models, compute and incentives, and it calls for rapid defensive adoption of these tools. The bulletin also highlights the need for stronger domestic and cross-border coordination, especially around vulnerability remediation, third-party risk and resilience planning.
Federal Reserve Board publishes AI buildout tracker, finds investment and adoption rising but broad labor effects limited
The Federal Reserve Board published a FEDS Note outlining public indicators to track generative AI's economic effects across capabilities and costs, firm investment and adoption, and productivity and labor. The note finds rapid capability gains, heavy investment and rising adoption, but says aggregate productivity and labor market effects remain limited and concentrated as of 2026.
Australian Securities & Investments Commission sets out four priorities for market innovation and capital formation after industry roundtable
The Australian Securities & Investments Commission published Report 835 and a summary of its market innovation roundtable, identifying four priorities for Australia’s capital markets. These were global competitiveness, stronger core infrastructure, innovation aimed at real market frictions and productivity, and investor protection through clear rules and safeguards. ASIC plans targeted industry workshops and engagement with regulatory peers before reconvening participants.
Organisation for Economic Co-operation and Development publishes paper on AI and Open Finance, highlighting synergies, trade-offs and amplified risks
The Organisation for Economic Co-operation and Development has published a paper on the interaction between AI and Open Finance in financial services. It says shared financial data can strengthen AI-driven personalisation, efficiency and inclusion, but can also intensify risks around bias, privacy, cybersecurity and consumer protection. The paper also flags trade-offs on consent, data minimisation and market structure, and points to stronger governance and cross-sector co-operation as key safeguards.
Czech National Bank research brief finds AI can affect 23.2% of Czech occupational tasks, with exposure concentrated in cities and information intensive sectors
The Czech National Bank published a research brief estimating that AI can affect 23.2% of tasks linked to Czech occupations. Exposure is concentrated in larger cities, finance and other information-intensive sectors, and among higher-wage and more educated workers. The brief says this uneven pattern could matter for inflation, household resilience and banks’ credit risk.
Federal Reserve Board note links the US AI buildout to stronger high tech imports prolonged current account deterioration and import price risks
The Federal Reserve Board published a FEDS Note saying the U.S. AI investment boom is likely to increase imports of high-technology equipment, widen the current account deficit and raise import price pressures. The analysis finds that investment-specific technology shocks have historically worsened the current account by about 10 percent relative to its average deficit, with potentially larger and longer-lasting effects now because AI supply chains are concentrated in East Asia and equipment prices have started to rise. Foreign spillovers are strongest for key technology suppliers such as Taiwan, Korea and Mexico.
Central Bank of the Philippines issues voluntary AI governance principles for supervised financial institutions
The Central Bank of the Philippines has issued voluntary AI governance principles for all BSP-supervised financial institutions. The STARS-based guidance sets minimum supervisory expectations for firms developing AI governance and risk management frameworks, with proportionate application based on the scale and risk of their AI use. It also covers outsourced providers involved in AI-related activities under a shared responsibility model.
Bank for International Settlements publishes working paper finding AI investment may exceed efficient levels by around 50% and heighten bust risks
The Bank for International Settlements published a working paper arguing that the AI build-out may be generating material over-investment and financial fragility. Its model estimates investment at about 50% above the efficient level under a conservative baseline, with debt and circular financing increasing the risk of fire sales and contagion. The paper also finds risks rise sharply as the build-out grows, with expected net economic surplus turning negative at around USD 3 trillion in investment.
Organisation for Economic Co operation and Development flags intensifying AI competition risks despite rapid price declines and new entry in foundation models
The Organisation for Economic Co-operation and Development said AI markets show a mixed competitive picture. Foundation models have become cheaper and more crowded, but the OECD warns that concentration in chips, cloud, data and skills, combined with first-mover advantages and vertical integration, could entrench a small number of firms. It calls for closer monitoring and policy action to preserve contestability across the AI value chain.
Dubai Financial Services Authority reports stronger audit inspection results and first transparency reports for Public Listed Company auditors
The Dubai Financial Services Authority's latest Audit Monitoring Report shows improved audit inspection results in the Dubai International Financial Centre and introduces the first transparency reports by auditors of Public Listed Companies under DFSA rules. The cycle covered 26 inspections across 93 files, while audit fees rose 74% to USD 33.5 million and the DFSA flagged engagement staffing and five recurring audit quality themes for further attention. Its 2026 inspection focus will include Systems of Quality Management, artificial intelligence in audit and revenue recognition.
Guernsey Financial Services Commission tells firms to review technology risk controls as AI speeds vulnerability discovery
The Guernsey Financial Services Commission has told regulated firms to review technology risk management after advances in AI increased the speed and scale of software vulnerability discovery. Its Dear CEO letter focuses on vulnerability management, patching and outsourced provider oversight. Boards and senior management are expected to ensure technology risk is monitored continuously and that remediation can be accelerated without weakening controls.
European Central Bank Banking Supervision directs significant institutions to submit AI cyber-risk action plans by October 31
European Central Bank Banking Supervision has directed significant institutions to assess AI-enabled cyber threats and submit action plans by October 31, 2026. Plans must address accelerated patching, exposed assets, monitoring, third-party risk and longer-term operational resilience. The ECB has extended the annual IT Risk Questionnaire deadline to February 2027 to support this work.
Dutch Authority for the Financial Markets publishes Financial Stability Committee warning on AI driven cyber risk and private credit transparency
The Dutch Authority for the Financial Markets published a Financial Stability Committee update warning that advanced AI models are increasing the urgency of stronger cyber resilience in finance. The committee called for faster vulnerability management and better coordination and information sharing across sectors. It also said private credit's rapid growth requires better data on exposures, credit quality and links to the wider financial system.
De Nederlandsche Bank reports Financial Stability Committee urges stronger cyber coordination and better private credit data
De Nederlandsche Bank said the Dutch Financial Stability Committee has warned that advanced AI models are changing cyber risk and that financial institutions need to adapt their risk management and resilience. The committee also said financial stability risks remain high and called for stronger cross-sector cyber coordination and information sharing. It separately flagged rapid private credit growth and said better data are needed on exposures, credit quality and interconnectedness.
European Systemic Risk Board warns frontier AI models could intensify systemic cyber risks in EU finance
The European Systemic Risk Board has warned that frontier AI models could raise systemic cyber risks for the EU financial system by helping threat actors find vulnerabilities and conduct attacks faster and at greater scale. It also flagged strategic dependency risks from the concentration of leading AI providers outside the EU. The ESRB will review the issue in its quarterly risk assessments and may take further action if needed.
European Banking Authority backs ESRB warning on frontier AI cyber risks and urges stronger financial sector defences
The European Banking Authority, alongside the other European Supervisory Authorities, backed the ESRB warning that frontier AI models are increasing systemic cyber risks for the financial sector. It urged financial entities to strengthen cybersecurity capabilities and called on supervisors to reflect the risks in their oversight. The authorities will continue monitoring the issue and clarify supervisory expectations under DORA.
Central Bank of the Philippines issues frontier AI cyber risk recommendations including hardware-backed MFA for privileged access
The Central Bank of the Philippines issued guidance for supervised institutions on managing cybersecurity risks from frontier AI systems that could automate vulnerability discovery and multi-stage attacks. It recommends stronger attack-surface visibility, zero trust and patching controls, hardware-backed MFA for privileged access with passwords and SMS or push methods discontinued for that purpose, AI-enabled defensive tools, business continuity reviews and institution-specific AI governance frameworks.
Federal Reserve Board publishes staff note on AI driven investment booms and overinvestment risk
The Federal Reserve Board published a staff FEDS Note arguing that major technologies such as AI can trigger long investment booms but can also end in overinvestment and excess capacity. The note says this can happen in a fully rational setting when investors learn about the scale of a technology shock through continued investment. It links the framework to the 1990s IT boom and to the recent AI-driven rise in U.S. IP and equipment investment.
OECD working paper finds AI could partly offset ageing-related growth drag but may be harder to deploy in older societies
The OECD has published a working paper arguing that AI could partly offset the economic drag from population ageing through productivity gains and relief for labour shortages. Using PIAAC data, it finds overall AI exposure peaks in mid-career, while younger workers face more automation exposure and older societies may struggle more with the reskilling, labour reallocation and business dynamism needed to capture AI gains. The paper also highlights risks for entry-level jobs and says more research is needed on the long-run productivity effects of AI.
Financial Conduct Authority publishes retail finance AI review with seven recommendations on agentic finance and supervision
The Financial Conduct Authority has published The Mills Review, concluding that AI will reshape retail financial services by 2030 through changes to firm operations, consumer journeys, competition and fraud and cyber risk. The report says about 11 million UK adults would likely use autonomous AI within pre-set goals and recommends seven actions, including adapting regulation, scaling the FCA's AI Lab and preparing for agentic finance and AI-enabled supervision. The FCA also plans an AI good and poor practice publication later this year.
Vietnam State Securities Commission joins review at Ho Chi Minh Stock Exchange to inform Securities Law amendments
The Vietnam State Securities Commission joined a review at the Ho Chi Minh Stock Exchange to assess implementation of the Securities Law and support planned amendments under the National Assembly's 2026 legislative program. Ho Chi Minh Stock Exchange proposed stronger use of artificial intelligence and big data in market surveillance, broader exchange-traded fund rules, and continued infrastructure and supervisory upgrades. Discussions also covered administrative simplification, new products and services, and a securities sandbox.