Federal Reserve Governor Christopher J. Waller said artificial intelligence (AI) could deliver significant efficiency and security gains in cross-border payments, and that "agentic commerce" — in which autonomous AI agents shop and pay on behalf of consumers and businesses — could significantly reshape commerce and payments if adoption scales.

Waller delivered the remarks in a speech titled "Payments in the Age of AI Agents" at Sibos 2026 in Miami, Florida, on September 29, 2026. Sibos, the annual global financial-industry conference organized by the banking cooperative SWIFT, draws banks, central banks, and technology providers from around the world. The full text was published on the Federal Reserve Board's website.

Within the financial sector, Waller said, the payments industry has long been at the forefront of using AI to improve operations. "There is a rich history of deploying machine learning to combat payments fraud," he said, adding that the industry "was also an early adopter of large language models (LLMs) for payment reconciliation and similar tasks. Now it is helping to build foundational infrastructure for AI agents to operate more broadly in the economy." These agents can plan and execute multistep processes using LLMs, enabling them to transact autonomously.

Noting that he had discussed a range of emerging technologies at last year's Sibos conference, Waller said this year's focus was more specific: the state of AI in payments and how the technology could change the way payment systems operate. He organized his remarks around two subjects — the implications of AI for the efficiency and security of cross-border payments, and the potentially transformative emergence of agentic commerce.

AI's Role in Cross-Border Payment Efficiency and Security

"Much has been said about how stablecoins can potentially improve cross-border payments," Waller said, before turning to what AI can contribute. Stablecoins are digital tokens designed to hold a steady value against a reference currency such as the U.S. dollar, a property central to the argument that they could make cross-border transfers faster and cheaper. Combating illicit finance, he said, is one of the clearest examples of where AI can drive efficiency and security gains — work that is critical in cross-border payments, where threat actors can exploit the seams between jurisdictions.

LLMs possess contextual awareness that can greatly improve the accuracy of sanctions screening and anti-money-laundering systems, Waller said, citing research demonstrating that the models can significantly reduce false-positive alerts for illicit activity. The study, "Can LLMs Improve Sanctions Screening in the Financial System? Evidence from a Fuzzy Matching Assessment," by Federal Reserve Board researchers Jeffrey S. Allen and Max S. S. Hatfield, was published in September 2025 in the Board's Finance and Economics Discussion Series. Fewer false positives, he noted, free up resources to examine more complex, high-risk cases and allow legitimate transactions to proceed with less friction. Given the velocity of modern payment systems, he said he expects LLMs to "augment, rather than replace, faster, more traditional anomaly detection methods.n
On cybersecurity, Waller acknowledged widespread concern that AI could increase the volume and sophistication of cyberattacks. "These concerns are well founded," he said, pointing to the near-weekly release of new models that push the boundaries of cybersecurity and to instances of AI agents escaping their test environments and breaching external systems. In the payments world, a successful cyberattack against highly interconnected payment infrastructure could significantly disrupt international commerce.

But AI is not only an offensive tool. Cyber professionals are optimistic about using AI for cyber threat detection and exposure management, among other important defensive tasks, Waller said, adding that AI can support the development of new, more secure software and help catch vulnerabilities before they are introduced. His remarks cited a 2025 article in the Journal of Financial Regulation by Iñaki Aldasoro, Sebastian Doerr, Leonardo Gambacorta, Sukhvir Notra, Tommaso Oliviero, and David Whyte on generative AI and cybersecurity in central banking.

Waller also highlighted a structural asymmetry in cyber defense: threat actors need to exploit only one key vulnerability, while payment system operators and service providers must defend a large attack surface. "That asymmetry is precisely why the industry must move deliberately to leverage AI in strengthening the safety of our trusted payment systems," he said.

AI could additionally facilitate the many layers of search and optimization required in sending money across borders, Waller said. Payment routing involves finding the path that best balances cost, speed, and reliability, while cross-border payment service providers operate complex foreign exchange conversion and liquidity mechanisms — deciding when and where to convert currency and whether to net offsetting flows. If presented with the right data and criteria, AI agents have the potential to excel at solving these complex optimization problems, he said, helping to improve the efficiency of cross-border payments.

Agentic Commerce: Early Days, Two Models

Waller then turned to agentic commerce — using AI agents to make e-commerce purchases — an emerging focus for the payments industry. He noted that at the beginning of the summer he hosted an industry roundtable at the Federal Reserve on the topic; materials from the May 26, 2026 meeting are available on the Board's website. Market participants broadly agree that agentic commerce is in an early phase, he said, but it could significantly reshape commerce and payments if adoption scales.

At a high level, Waller described two primary models. In the agent-assisted model, a buyer uses an AI agent primarily for product search and discovery, but the buyer remains in control — making the decisions and handling the payment. In the agent-delegated model, a buyer grants an AI agent authority to shop and make payments on their behalf; the buyer may specify constraints and set up guardrails, but the agent operates autonomously. As an example, he described a buyer giving an agent access to a pre-funded virtual card with instructions to shop for groceries online based on past purchases.

Consumer-to-business transactions are seen as the first wave of agentic commerce, with agent-assisted use cases preceding more autonomous, agent-delegated ones, Waller said. Because delegation introduces higher risks of unintended purchases, the delegated model requires a more extensive buildout of trust mechanisms and guardrails.

Consumers are likely to access agents in two common ways: through general-purpose agents offered by AI firms and through merchant-specific agents. As personal AI assistants gain traction, they too could become a natural fit for agentic commerce.

For agents to be useful to consumers, they need broad, real-time knowledge of merchant inventories. Large online retailers are sharing inventory data in different ways, Waller noted: some work directly with AI firms to make their product catalogs available to third-party agents, while others surface catalog data through their own shopping agents. E-commerce platforms, meanwhile, are building tools to standardize digital product catalogs for small and medium-sized merchants and to make them discoverable by leading agent interfaces.

B2B Purchases and Machine-to-Machine Micropayments

While consumer shopping receives much of the attention, Waller said business-to-business (B2B) purchases may be especially well suited for agentic commerce. These purchases are often recurring and governed by rules — such as approved suppliers and budget limits — that create natural guidelines for agents. Agents could also potentially negotiate terms with suppliers and develop payment strategies to optimize working capital.

However, the higher transaction values typical of B2B commerce amplify the financial exposure from agent errors or unauthorized actions, so robust controls and monitoring systems will be essential, he said. B2B agents would also need to execute payments across a diverse set of rails: whereas payment cards dominate consumer e-commerce, businesses rely on a wider range of methods, including ACH, wire, instant payments, and cards.

Across these shopping contexts, agents will need to make micropayments along the way, Waller said. Before completing a purchase, an agent might pay for LLM queries, price data feeds, and other API calls. Small machine-to-machine payments are not new, but they could reach a much larger scale if shopping agents work around the clock at machine speed. Micropayments like these favor payment rails with lower flat fees, he noted.

Trust: Authentication, Liability, and Fraud

The biggest barrier to scaling agentic commerce — particularly the agent-delegated model — is building sufficient trust among buyers and sellers, Waller said, identifying three concrete challenges: authentication, liability, and fraud.

Agentic commerce transforms the payment authentication paradigm, he said: the question shifts from proving that a buyer is an authorized payer to proving that an agent has the authority to pay on the buyer's behalf. Capturing that distinction will require new authentication approaches.

Liability, he said, comes down to a simple question: "Who is on the hook if an agent makes the wrong purchase?" Current e-commerce liability frameworks — including network rules, consumer protection standards, and other regulations — could potentially be adapted for agentic commerce, but there may also be an opportunity to test and refine new approaches. For example, technical standards could be designed to give all parties a better understanding of what the buyer intended and how the agent carried it out, reducing the ambiguities that complicate many transaction disputes today.

Fraud poses a further risk to confidence in agentic commerce. Existing fraud detection and prevention systems are calibrated to human behavior and may not translate well to agents, Waller said, so fraud models and rules will need to be recalibrated to account for agent payment patterns.

Standards and Market Structure Questions

Market participants are actively developing standards to address these trust challenges and related issues, Waller said. Tech firms, e-commerce platforms, and payment service providers are building foundational protocols that standardize how agents interact with merchants and make payments. Card networks are publishing technical specifications that enable AI agents to make secure payments on a payer's behalf, including methods for registering agents and logging cardholders' approvals to conduct agentic payments.

A key question, he said, is whether agentic commerce will migrate toward platform-specific or interoperable standards. Platform-specific standards are designed around a single ecosystem; interoperable standards are designed to work across e-commerce systems, agent interfaces, and payment methods — and could help level the playing field for smaller merchants and payment providers seeking to operate in the space.

A related consideration is open versus closed agentic commerce systems, which share similarities with open- and closed-loop payment networks. An open system allows a range of shopping agents on its platform, while a closed system requires the use of a specific AI agent — for example, an online retailer that requires shoppers to use its proprietary agent and blocks others. Open systems support buyer choice and an active environment for developing new agentic commerce solutions, while closed systems give online retailers more control over their platforms and lead to more vertical integration. At this early stage, it is not clear which model have the edge in shaping agentic commerce, Waller said, but whichever gains traction could significantly influence how the market structure evolves.

Waller also flagged unresolved issues for the industry as the new mode of transacting matures: which standards are still missing for agents to carry identity, consent, and payment credentials across the full e-commerce stack; the right balance between user convenience and purposeful friction in agent authorization; and whether payment rails will need to adapt in a future where agentic payments scale dramatically. "These are questions worth pondering," he said.

Looking Ahead

Waller said he is keenly focused on how new technologies can affect the structure and functioning of payment systems globally. While his comments focused on within cross-border and e-commerce payments, AI agents may become integrated into a wide range of payment flows, and agentic transactions could materially change the frequency and timing of payments, as well as other characteristics.

"Meeting this moment requires a proactive approach that balances innovation with the safety, integrity, and stability that underpin trust in payments," he said, adding that he is very interested to see how the payments industry adapts to the use of AI in payments. For observers tracking the shift, the near-term markers will be whether platform-specific or interoperable standards gain adoption, how open and closed agent systems compete, and whether the industry fills in the pieces Waller flagged as missing — standards for carrying identity, consent, and payment credentials across the e-commerce stack.

Source and References

• Speech: Christopher J. Waller, "Payments in the Age of AI Agents," delivered at Sibos 2026, Miami, Florida, September 29, 2026. Full text: https://www.federalreserve.gov/newsevents/speech/waller20260928a.htm
• Prior speeches cited: "The Next Frontier of Payments Innovation" (Sibos 2025, Frankfurt, Germany, September 29, 2025); "Technological Advancements in Payments" (Wyoming Blockchain Symposium 2025, Teton Village, Wyoming, August 20, 2025); "Reflections on a Maturing Stablecoin Market" (A Very Stable Conference, San Francisco, California, February 12, 2025).
• Jeffrey S. Allen and Max S. S. Hatfield (2025), "Can LLMs Improve Sanctions Screening in the Financial System? Evidence from a Fuzzy Matching Assessment," Finance and Economics Discussion Series 2025-092, Board of Governors of the Federal Reserve System, September.
• Iñaki Aldasoro, Sebastian Doerr, Leonardo Gambacorta, Sukhvir Notra, Tommaso Oliviero, and David Whyte (2025), "Generative Artificial Intelligence and Cyber Security in Central Banking," Journal of Financial Regulation, vol. 11 (April), pp. 119–28.
• Roundtable materials: https://www.federalreserve.gov/paymentsystems/files/payments-innovation-roundtable-meeting-20260526.pdf (May 26, 2026 payments innovation roundtable, Federal Reserve Board website).
• Per the published text, Waller thanked Jeff Allen and Micah Gertzog of the Federal Reserve Board for assistance in preparing the speech; the views expressed are his own and not necessarily those of his colleagues on the Federal Reserve Board or the Federal Open Market Committee.