When a senior member of the Federal Reserve System's Board of Governors takes the stage at the world's premier financial-industry gathering to speak about artificial intelligence and payments, the message carries institutional weight that the sector cannot afford to dismiss. That is precisely what happened on September 29, 2026, when Governor Christopher J. Waller addressed delegates at Sibos 2026 in Miami, Florida, delivering a keynote framed around a deceptively simple question: what happens to payments when artificial intelligence agents, rather than human beings, become the primary initiators and authorisers of financial transactions?
The speech, subsequently indexed in the official speeches repository of the Bank for International Settlements, signals something of a watershed moment. Central bank officials have spent years discussing digital currencies, real-time rails, and cross-border interoperability. Waller's address at Sibos 2026 moves the conversation decisively into new territory: an era in which autonomous software systems — AI agents capable of perceiving their environment, making decisions, and executing actions without moment-to-moment human direction — are becoming active participants in the payments stack rather than passive analytical tools sitting alongside it.
Why Sibos, Why Now
Sibos, organised annually by SWIFT, draws together central bankers, commercial bank executives, payment infrastructure operators, and technology vendors in a single forum precisely because payments are a domain where policy, technology, and commerce are inextricably intertwined. Choosing this platform to deliver a speech on AI agents was itself a signal. Miami's 2026 edition of the conference arrives at a moment when the deployment of large-language models and autonomous agent frameworks has accelerated far beyond the proof-of-concept phase, with financial institutions across the globe integrating agentic systems into treasury operations, fraud detection, and increasingly, transaction execution. Governor Waller's decision to address the topic directly, under the BIS speech banner, underscores that regulators are no longer willing to treat AI agency in payments as a purely commercial question.
The Structural Challenge AI Agents Pose to Payments
Traditional payments architecture was designed with a human being at the origin of every transaction: a cardholder tapping a terminal, a corporate treasurer approving a wire, a consumer authorising a direct debit. Compliance frameworks, liability rules, authentication standards, and dispute-resolution mechanisms all presuppose that moment of human intent. AI agents dissolve that presupposition. An autonomous agent instructed to optimise a corporate liquidity position might initiate dozens of interbank transfers in a single second, each one technically compliant in isolation but collectively producing systemic patterns that no human operator consciously directed. The question Governor Waller implicitly raises is whether the current payments governance architecture — built on the assumption of traceable human decision-making — is fit for this new operating reality.
This is not a theoretical concern. Across the industry, treasury management systems are already deploying agent-like automation. Retail fintech platforms are experimenting with AI that can negotiate, book, and pay for services on a user's behalf. In wholesale markets, algorithmic systems that blur the line between analysis and autonomous action have been standard for years. What has changed is the generalisation of these capabilities: where once agentic behaviour was confined to narrow, well-defined tasks, modern AI agents can operate across open-ended domains, adapting to novel circumstances without explicit human instruction at each step.
Regulatory Legibility and the Identity Problem
One of the most consequential dimensions of the AI-agent challenge in payments is identity. Anti-money laundering frameworks, know-your-customer requirements, and sanctions screening all rest on the ability to identify who is transacting. When an AI agent executes a payment, the legal and regulatory question of "who" is acting becomes genuinely complex. Is it the individual or institution that deployed the agent? The developer who built it? The model that powers it? Governor Waller's engagement with this topic from a Federal Reserve platform suggests that the central banking community is beginning to formalise its thinking on agent identity and accountability — a development that will have direct consequences for compliance teams, correspondent banks, and infrastructure providers worldwide.
The BIS publication of the speech further amplifies its reach, ensuring that central banks and supervisory authorities across member jurisdictions receive the Federal Reserve's framing as a reference point. In the absence of a globally harmonised standard for AI agent conduct in financial transactions, a speech of this nature from a G10 central bank governor carries de facto norm-setting influence.
What This Means for the Industry
For financial institutions, the implications are immediate and multilayered. Compliance functions will need to begin mapping their existing authentication, authorisation, and audit-trail frameworks against scenarios in which the transacting entity is an AI agent rather than a human principal. Payment infrastructure operators — from card networks to real-time gross settlement system operators — will face questions about how their rule books apply when the party initiating a transaction cannot provide biometric authentication or sign a legal agreement. Fintech developers building agent-based financial tools will need to anticipate regulatory requirements around agent disclosure, liability assignment, and transaction limits that do not yet formally exist but are clearly coming into view.
Governor Waller's Sibos address does not represent the end of the regulatory conversation around AI in payments. It represents, more accurately, the moment when that conversation moved from the industry's innovation labs into the official monetary policy and financial stability discourse. The payments ecosystem has navigated transitions before — from paper to electronic, from batch to real-time, from domestic to cross-border. The transition to an age of AI agents may prove the most structurally demanding of all, precisely because it challenges not just the technology of payments but the foundational assumptions about human agency on which the entire governance architecture was built.
Written by the editorial team — independent journalism powered by Codego Press.