The rapid advancement of artificial intelligence has outpaced regulatory frameworks designed to oversee financial systems. At a recent United Nations panel on the Future of Money, Dottie Romo, Chief Risk and Control Officer at the US Internal Revenue Service, outlined the challenge facing financial regulators as AI systems gain control over payments at scale.
Research tracking AI agent payments between July 23 and August 26 documented 6.4 million transactions carrying $119,947 across blockchain networks Base and Solana. While individual transactions remain small—with 90.8% worth less than one cent—the volume reveals the speed at which machine-driven commerce operates. The same AI agent has recorded nearly 200 million settlement transactions since launch.
Romo highlighted the fundamental mismatch between regulatory processes and AI capabilities. "They're making millions of decisions in minutes," she said, pointing out that current oversight relies heavily on periodic reports to detect fraud and control failures. This approach becomes impractical when systems operate at machine speed.
The solution, Romo suggested, may require a counterintuitive approach: algorithms monitoring other algorithms. "We're not going to be able to do that in a fast enough pace" through traditional human oversight, she noted, calling for more real-time market monitoring while maintaining human involvement in critical decisions.
Julius Moye, Manager at Mastercard's Financial Crime Solutions, referenced past automated system failures as cautionary examples, including the 2012 Knight Capital trading disaster. He proposed a tiered response model where machines detect anomalies and automatically contain problems, with human intervention reserved for serious incidents.
Dino Cataldo Dell'Accio of the UN Joint Staff Pension Fund stressed that automation cannot eliminate accountability. He argued that responsibility must ultimately trace back through three questions: who developed the code, who implemented it, and who oversees it.
The practical reality emerging from these discussions is that regulators face a structural challenge they cannot solve through conventional means. As AI systems continue processing millions of transactions monthly, financial oversight may require machine-based supervision rather than human-centered regulation.
