Payments In The Age Of AI Agents

TL;DR

Prime Big Deal Days · Oct 6–7Offer from Amazon

Get smart everyday buys delivered free — and shop member deals

• Fast, free delivery on millions of items
• Access to Prime Big Deal Days deals on October 6–7
• Prime Video, Amazon Music and more included

Start your free Prime trial →Free trial for eligible customers · Cancel anytime

As an affiliate, we earn on qualifying purchases.Hide

Federal Reserve Governor Christopher Waller said AI could help make cross-border payments more efficient and secure, and could support a shift toward purchases made by autonomous agents. He described agentic commerce as being at an early stage, with delegated purchases requiring stronger trust mechanisms and safeguards. The speech did not set out a policy proposal or timetable for adoption.

Federal Reserve Governor Christopher Waller said AI could reshape payments, from cross-border transaction screening and routing to online purchases made by autonomous agents, in a speech at Sibos 2026 in Miami on September 29. He described agentic commerce as an early-stage development and said its more autonomous forms would need stronger safeguards because they give software authority to spend on a buyer’s behalf.

Waller said large language models could improve sanctions screening and anti-money-laundering systems by reducing false-positive alerts, citing research without detailing the studies in the speech. Fewer false alerts could free staff to examine more complex cases and reduce delays for legitimate payments. He said these models would likely augment faster, traditional anomaly detection, rather than replace it.

AI could also help payment providers choose routes that balance cost, speed and reliability, and manage foreign-exchange conversion and liquidity. Waller presented these as potential applications, dependent on systems receiving suitable data and criteria. He also warned that AI may raise the volume and sophistication of cyberattacks, while offering defensive uses such as threat detection and identifying software vulnerabilities.

Waller distinguished two types of agentic commerce. In agent-assisted shopping, an agent searches for products while the buyer decides and pays. In agent-delegated shopping, the buyer authorizes an agent to purchase within set constraints, such as using a pre-funded virtual card to order groceries based on past purchases. He said business-to-consumer purchases are expected to be an early use case, with assisted shopping preceding more autonomous delegation.

Trust Will Shape Agent Purchases

The distinction between an agent that recommends products and one that can spend money changes who acts at checkout. Delegation may reduce the work involved in routine purchases, but it also creates questions about authorization, spending limits and unintended transactions. Waller said delegated commerce brings higher risks and calls for a more extensive buildout of trust mechanisms and guardrails.

For payment providers, AI could affect both the front end of shopping and the infrastructure behind a transaction. Automated screening and routing may improve the handling of cross-border payments, while highly connected payment systems could be exposed to cyberattacks. Waller stressed that defenders must protect broad systems even as attackers may need to exploit only one vulnerability.

From Fraud Screening to Agents

Waller placed the new applications within a longer history of AI use in payments. He said the industry has used machine learning to combat fraud and adopted large language models for tasks such as payment reconciliation. The next step, in his account, is infrastructure that lets agents plan and carry out multistep processes, including transactions.

He said he had hosted an industry roundtable on agentic commerce at the Federal Reserve earlier in the summer. Participants broadly agreed the field remained at an early stage, according to his speech. The remarks were delivered at Sibos, a payments and financial services conference, and focused on possible developments rather than a formal Federal Reserve rule or program.

“I expect LLMs to augment, rather than replace, faster, more traditional anomaly detection methods.”

— Christopher Waller, Federal Reserve governor

Adoption and Safeguards Remain Open

Waller did not provide an adoption timetable, estimate how widely agentic commerce is being used, or specify common standards for authorizing agents and limiting their spending. It is also unclear which companies or payment systems will supply the trust mechanisms he said delegated purchases require.

The speech cited research on false-positive reductions in illicit-finance screening but gave no figures or study details. It did not quantify expected changes in payment costs, speed or reliability, and the potential benefits of AI-enabled cyber defense remain uncertain alongside the risks Waller identified.

Industry Work Moves to Guardrails

Waller’s speech points to further industry work on the infrastructure and safeguards needed for agents to transact. He did not announce a specific Federal Reserve action or a next policy milestone. The practical questions for market participants include how consumers grant and revoke authority, how agents stay within spending constraints, and who is responsible when an agent makes an unintended purchase.

For cross-border payments, providers will need to evaluate AI tools against operational needs, data quality and cybersecurity risks. Waller called for the industry to move deliberately as it uses AI to strengthen payment system safety; the timing and form of any wider adoption remain unsettled.

Key Questions

What is agentic commerce?

Agentic commerce is the use of AI agents to help make online purchases. In delegated commerce, a consumer gives an agent authority to shop and pay within specified constraints.

How does agent-assisted shopping differ from delegated shopping?

With agent-assisted shopping, the agent helps with product search while the buyer decides and pays. With delegated shopping, the buyer authorizes the agent to make purchases on their behalf, subject to any guardrails they set.

How could AI affect cross-border payments?

Waller said AI could support sanctions and anti-money-laundering screening, help route payments by cost, speed and reliability, and assist with foreign-exchange and liquidity decisions. These are potential uses; the speech did not quantify their effects.

What risks did Waller identify?

He warned that AI could make cyberattacks more frequent or sophisticated and said delegated shopping carries a higher risk of unintended purchases. He argued that more autonomous commerce requires stronger trust mechanisms and safeguards.

Source: primary