Swiss Financial Market Supervisory Authority Chair Marlene Amstad outlined in a speech how the regulator is using supervisory technology to improve document reviews, market abuse detection and digital asset oversight. She emphasized that AI tools should augment rather than replace supervisors, with human staff retaining judgment and accountability, while identifying speed, reliance on a small number of model providers and governance as key risks from AI adoption in finance. FINMA is testing a locally hosted generative AI tool that flags potentially relevant passages before on-site inspections, with a second model checking suggestions for hallucinations. It is also developing supervised machine learning and network analysis tools to identify suspicious trading and trace flows of inside information. A crypto-monitoring dashboard already combines quarterly reported holdings with daily token prices to provide near real-time views of institutional and systemwide exposures, supporting early warnings for concentration and blockchain-related operational risks. FINMA plans to add features and extend the dashboard to all supervised licence types over the medium term. Separately, an International Organization of Securities Commissions initiative chaired by FINMA has launched the first phase of developing crypto-market supervisory tools, with the build phase scheduled for September. An IOSCO survey cited in the speech found that 91% of authorities rate efficiency and effectiveness as high or critical SupTech priorities, while only 4% prioritize reducing headcount.

Original source Back to tracker