Governor Christopher J. Waller spoke Tue 29 Sep 2026 at Sibos 2026 in Miami, Florida.

The payments industry has long used machine learning against fraud and was an early adopter of large language models for payment reconciliation. Those agents can plan and execute multistep processes and transact autonomously.

Agentic commerce has two models. In the agent-assisted model, the buyer stays in control of the decision and the payment. In the agent-delegated model, the buyer grants an agent authority to shop and pay on their behalf.

The biggest barrier to scaling agentic commerce, particularly the agent-delegated model, is trust: authentication, liability, and fraud. Authentication shifts from proving a buyer is an authorized payer to proving an agent has authority to pay on the buyer's behalf.

Before a purchase, an agent may pay for language-model queries, price data, and other API calls. Micropayments of that kind favor payment rails with lower flat fees.

Meeting this moment requires a proactive approach that balances innovation with the safety, integrity, and stability that underpin trust in payments.

As-of Tue 29 Sep 2026. Federal Reserve.