The rapid expansion of autonomous artificial-intelligence systems into finance is forcing regulators to confront a fundamental question: who is responsible when an AI agent makes a financial decision?
Bank of England Governor Andrew Bailey raised the issue during the Group of Thirty’s international banking seminar in Bangkok, warning that users could misunderstand where legal responsibility sits when AI agents are authorized to make payments or conduct other financial transactions.
The concern extends beyond conventional software risk.
AI agents can increasingly execute tasks rather than simply provide information. In finance, that could eventually include initiating payments, trading assets, interacting with financial institutions or making other decisions involving real money.
Bailey has previously highlighted the legal problem that can emerge when autonomous software acts beyond the instructions originally provided by its user. His latest remarks place that question within a broader financial-stability debate as adoption accelerates.
Senior policymakers participating in the Bangkok discussions emphasized stronger model testing, anti-fraud controls and continued reliance on established financial safeguards, including capital and leverage requirements.
The regulatory challenge is significant because financial law generally assumes that identifiable people or legal entities ultimately carry responsibility for financial decisions. Autonomous AI systems do not themselves have conventional legal personality.
Financial institutions deploying agents may therefore need governance systems capable of establishing what an AI system was authorized to do, what it actually did and who remains accountable for the outcome.
That requirement could become increasingly important as financial AI moves from analysis and recommendations toward execution.
The next stage will be whether regulators translate those concerns into specific requirements governing authorization, model testing, transaction limits, audit trails and liability.
