This report examines agentic artificial intelligence deployment across financial services and the fundamental gaps in authentication, liability allocation, and governance infrastructure that threaten safety and sustainability. Based on research with financial institutions deploying agents both internally and for agentic commerce, the findings show that traditional fraud authentication frameworks are fracturing because agents operate without behavioral deviation, shared infrastructure for real-time agent authentication does not exist, and unresolved liability allocation in Regulations E and Z is blocking institutional investment in protective systems. Governance and regulatory uncertainty rank as greater constraints than technology, talent, or cost — yet adoption speed is far outpacing governance maturity. Nearly all institutions pilot agentic AI in payment fraud detection despite lacking formal policies, regulatory exposure assessments, or reliable agent activity detection. The report identifies four critical gaps requiring immediate action from financial institutions, regulators, and technology providers to close before widespread disputes, enforcement, or market fragmentation force costly remediation.
Clients of Datos Insights’ Fraud & AML practice may access this report.
About the Author
Other Authors
