Global Banks set ground rules for Agentic Commerce as Fraud and Liability risks rise

Six major banks have joined forces to establish principles for trusted agentic commerce, warning that the rapid adoption of AI shopping agents could outpace existing safeguards around fraud, scams, privacy and payment liability.

Global Banks set ground rules for AI Commerce

Bank of America, NatWest, ING, Capital One, Commonwealth Bank of Australia and ASB Bank have published a joint framework covering five areas: transparency, safety, privacy and data, consumer choice and interoperability.

The initiative reflects a growing concern across the payments industry: AI agents are becoming capable of selecting products, choosing payment methods and completing transactions, but consumers remain unsure who is responsible when something goes wrong.

Trust Is Lagging Behind the Technology

The trust gap is significant.

Visa research published this month found that only 23% of US consumers trust generative AI to handle payment transactions on their behalf, although that figure rises to 61% when Visa is responsible for securing the transaction.

That distinction matters because agentic commerce changes the traditional payments relationship.

Instead of a consumer directly selecting a merchant and authorising a payment, an AI system may increasingly make recommendations, choose how to pay and execute the transaction within parameters set in advance.

The banks warn that consumers need clarity over whether an agent is acting in their interests, how much authority it has and who is accountable if it buys the wrong item, exceeds a spending limit or falls victim to fraud.

Fraud and Liability Become Harder to Untangle

For merchants and payment providers, the challenge is equally significant.

AI agents may introduce new uncertainty around chargebacks and disputes if a customer later argues that a transaction was technically authorised but not what they intended. Banks are also concerned that agents could expose card credentials, mishandle sensitive data or steer customers towards payment methods offering weaker protections.

That makes identity, consent and liability central to the next phase of payments infrastructure.

Bank of America says building confidence will require clear approaches to authorisation, fraud prevention, liability management and customer protection.

Agentic Commerce Is Already Reaching Retail

This is not simply a theoretical concern.

John Lewis recently said searches originating from AI agents had risen to 2.5% of product searches from 0.3% a year earlier, suggesting that AI-led shopping behaviour is already becoming commercially relevant.

The banks now plan to work with policymakers and other industry participants on practical implementation frameworks.

The direction of travel is clear. Agentic commerce will not scale on convenience alone.

The payments industry now has to prove that AI can be given greater authority without weakening the protections, accountability and trust on which digital commerce depends.