Financial institutions are already putting AI agents to work: executing trades within mandates, triaging compliance alerts, and settling value on tokenized rails. No major jurisdiction has yet written rules specifically for agentic finance. Instead, institutions must stitch compliance together from frameworks built for other purposes, including the EU AI Act, MiCA, DORA, the ESMA DLT Pilot Regime, market-conduct and market-abuse rules, and model risk guidance in the US.
That leaves boards, risk committees, and regulators asking the same question: who is accountable when an agent acts? Answering it requires fluency in AI governance, securities regulation, and market infrastructure at the same time, across more than one jurisdiction.
And every major market is answering it differently. Europe is legislating comprehensively. The US is working through agency guidance and supervisory signals. Asia's financial centres and fast-moving markets like Brazil are building their own approaches in parallel. A firm running the same agent, token, or settlement rail in several of these markets faces different expectations for the same system.
My work is building one governance and regulatory approach that holds up in each of them, drawing on firsthand experience across Europe, the United States, Asia, and Latin America: from structuring transactions in London, Paris, and Tokyo, to advising finance ministries in Japan and South Korea, to leading a regulated DLT platform's European licensing and US expansion.
