Banks Pilot AI Agent Payments as Regulators Shift Focus to Accountability
Financial institutions are testing autonomous AI transactions while regulators warn that humans remain responsible for software-driven spending.
The financial industry is moving toward a future where AI agents do more than offer advice—they are beginning to execute payments. This transition is forcing a fundamental rethink of how the global payment system maintains security and accountability when software, rather than a human, initiates a transaction.
Recent live demonstrations prove that agentic payments can function within existing regulatory frameworks. In 2026, Santander and Mastercard executed Europe's first live payment initiated by an AI agent within a regulated banking environment. Similarly, Nordea and Mastercard successfully completed a live AI-agent transaction in Finland. These pilots utilize tokenized credentials and pre-authorized permissions to ensure that autonomous spending remains within strict boundaries.
The Accountability Gap
Traditional financial systems assume every payment instruction links to a person or entity with clear, verifiable authority. Agentic AI introduces a layer between the human user and the transaction, complicating identity-based controls. This shift increases the risk of rapid, automated errors or fraudulent requests executed at scale before a human can intervene.
To address this, regulators are clarifying that delegating a task to an AI does not equal a delegation of responsibility. In March 2026, the UK Competition and Markets Authority (CMA) issued guidance stating that organizations remain fully accountable for the actions of AI agents acting on their behalf. Similarly, the EU AI Act mandates strict transparency, human oversight, and record-keeping for high-risk AI systems, including those used in critical financial services such as credit scoring.
Redefining Financial Trust
As AI agents gain the ability to move money autonomously, the industry must shift from simple identity verification to a model based on intent, authority, and context. The World Economic Forum notes that the central challenge is not whether AI agents should be used in finance, but how institutions can support their use without weakening accountability.
Failure to establish clear audit trails and explainability for AI-driven decisions could undermine trust in the global payment system. Without these safeguards, the industry risks creating new vulnerabilities for systemic error and sophisticated fraud, where the speed of AI execution outpaces the speed of human oversight.
The Path Forward
Industry leaders are now watching how broad agentic commerce programs, such as Visa Intelligent Commerce, will scale across different jurisdictions. The next phase of adoption depends on the development of standardized protocols for 'intent verification'—ensuring an agent acts on a legitimate human request. While the technical ability to move money via AI has been demonstrated, the legal and operational frameworks for handling disputed AI transactions remain a critical area of development.