Six global banks, including NatWest, Bank of America and ING, have published a shared set of principles for AI-powered shopping and payments, calling for stronger safeguards as so-called agentic commerce develops.
The joint announcement from NatWest Group was published on 22 September 2026 with ASB Bank, Bank of America, Capital One, Commonwealth Bank of Australia and ING Group. The banks say AI agents could increasingly help people choose products and services and make payments on their behalf, but that trust will depend on customers retaining control and understanding what the technology is doing.
Five principles for AI-assisted shopping
The banks have organised their framework around five areas: transparency, safety, privacy and data, choice, and interoperability. They want customers and merchants to be able to understand when an AI agent is involved, keep meaningful control over payments and choose which services they use.
The principles are not new regulations and they do not create legal obligations for retailers. Instead, the six banks are presenting them as a starting point for wider industry discussion. NatWest says the group plans to develop a further paper setting out how the principles could be put into practice.
A Reuters report on the announcement highlighted concerns around scams, fraud and data privacy. Among the risks identified are AI agents requesting customers' card details and entering them directly into websites, or directing shoppers towards payment methods that may offer weaker protections.
AI is already becoming part of product discovery
The announcement arrives as AI tools are moving beyond answering questions and towards helping people research, compare and buy products. That matters to retailers because the customer journey may increasingly begin inside an AI service rather than on a search engine, marketplace or retailer's own website.
There are early signs of that shift in the UK. Reuters reported that John Lewis said in September that searches originating from AI agents had risen from 0.3% to 2.5% over the previous year, with the trend accelerating. The figure does not show that AI agents are already completing a comparable share of purchases, but it does indicate that automated product discovery is becoming a more visible source of customer activity.
Roled recently covered Google's introduction of AI Search visibility reporting, which gives website owners more information about when their pages appear in Google's generative AI features. The banks' new principles look at a later stage of the same broad change: what happens when an AI system moves from helping someone discover information to assisting with a transaction.
What businesses may need to watch
For online retailers, the immediate issue is not to redesign checkout around AI agents. The technology, standards and consumer protections are still developing, and the banks themselves describe their principles as a foundation for further work.
What is becoming clearer is that businesses may eventually need to think about automated customers as well as human ones. Product information, pricing, availability, payment choices, privacy notices and returns policies may be read or interpreted by AI systems acting for shoppers. If those systems begin taking more direct action, questions around authentication, consent, fraud checks and responsibility for mistakes become more important.
The banks' intervention is therefore useful less as a prediction of how people will shop and more as evidence that the payments industry is preparing for the possibility. Their next paper, which is expected to look at implementation, should provide a clearer indication of how these principles could translate into practical requirements for merchants, payment providers and AI platforms.
