Banks say AI shopping agents are outpacing fraud protections

According to Mediacorp, a group of banks has told policymakers that AI shopping agents are moving faster than the rules meant to protect the people who use them. NatWest, Bank of America, ING, New Zealand's ASB Bank and US lender Capital One are among the banks behind the report.

The banks are not warning about a hypothetical. Tech companies including OpenAI, Anthropic, Google and Meta are already building AI chatbots that shop on a customer's behalf, choosing products and completing purchases. Retailers are already adjusting for it. John Lewis says searches arriving from AI agents rose from 0.3 per cent to 2.5 per cent in a year, and the trend is accelerating.

The gap the banks describe

Standards and consumer protections have not kept pace with the technology. That is the banks' own framing, not an inference. Customers want the convenience. They are enthusiastic about letting an AI agent handle shopping. But the report says those same customers are unclear whether the AI will act in their interests, and unsure who to contact if something goes wrong.

Two specific risks stand out in the report. One is agents that request a customer's card details and enter them directly into a website, which hands sensitive payment data to a system the customer cannot fully see or control. The other is agents steered toward payment methods that carry weaker consumer protections than the ones a person would ordinarily choose.

The sequence matters here. Retailers are already optimising for AI-driven traffic, and adoption is accelerating, while the rules for disclosure, liability and data handling are still being drafted. The industry is building the sales channel before it has agreed who is responsible when that channel is exploited by a scam.

What the banks are proposing

The banks plan to raise a set of proposals with policymakers. These include a requirement to disclose when an AI agent is involved in a transaction, so a customer knows a bot acted on their behalf rather than assuming a human did. They want greater transparency over how an AI agent reaches its decisions. They want safeguards around customer data. And they want consumers and merchants to be free to choose which AI-powered shopping service they use, with different systems able to work together rather than locking customers into one provider.

None of this is law yet. It is a set of principles the banks intend to discuss with policymakers, at a point when agentic shopping is already live and growing. The report itself says the technology is ahead of the protections built around it. Anyone using an AI shopping agent today is doing so inside that gap.