The International Organization of Securities Commissions has published a Supervisory Toolkit for AI Use in Capital Markets, giving securities regulators a practical, non-binding framework for supervising AI systems used by regulated firms. The toolkit is intended to support risk-based and proportionate oversight across the full AI lifecycle and applies to traditional machine learning models, generative AI and emerging agentic AI techniques. It forms part of IOSCO’s phased work on regulatory and supervisory responses to AI-related risks in capital markets, including risks to investor protection, market integrity and financial stability. The toolkit is structured around three layers. It identifies risk areas for supervisory consideration, sets out oversight tools for governance and risk management, third-party and outsourcing risk, disclosure, and recordkeeping and reporting, and provides indicators and engagement methods for monitoring AI adoption and use. IOSCO has also issued a standalone extract for direct use in supervisory work such as on-site examinations and inspections. The report notes that wider AI use in investment processes, risk management and operational functions can increase complexity, reduce transparency and deepen third-party dependencies, and it flags AI-enabled cyber capabilities as a growing source of supervisory concern. IOSCO said its next phase of work will review emerging industry practices in governance, disclosure, and recordkeeping and reporting for AI systems in capital markets. A survey included with the report is open until 26 June to inform that review, and IOSCO said it will continue coordinating on AI developments with other international bodies, including the Financial Stability Board.