AI agents are getting closer to choosing products and completing purchases for consumers — and major banks are now asking what happens when the software buys the wrong thing, spends too much or sends money into a scam.

WHAT’S HAPPENING

Banks including Bank of America, NatWest, ING, Capital One, ASB Bank and Commonwealth Bank of Australia are calling for safeguards around agentic commerce — AI systems that can move beyond recommending products and actually make purchases for users.

The banks say the technology is developing faster than some existing consumer protections and industry standards. Their concerns include AI agents handling payment information, selecting the wrong products, overspending or directing consumers toward payment methods with weaker protections.

WHY IT MATTERS

AI shopping changes the role of the chatbot.

Instead of simply telling you what to buy, the AI may eventually be able to search, compare, decide and pay within limits established by the user.

That creates a new question:

When an AI makes the purchase, who is responsible when something goes wrong?

WHO BENEFITS

Consumers could save time by allowing AI to search prices, compare products and complete routine purchases automatically.

Retailers, payment companies and AI developers could gain access to a new layer of automated commerce.

WHO LOSES

Consumers could face new risks if an AI agent makes an incorrect purchase, exposes payment information or interacts with a fraudulent seller.

Retailers could also lose some control over the customer relationship if AI agents increasingly decide where consumers shop.

WHAT HAPPENS NEXT

Banks are proposing discussions around disclosure when AI participates in a transaction, greater transparency into how agents make decisions and stronger protections for customer data.

The bigger shift is already becoming visible:

AI is moving from helping people decide what to buy to potentially being trusted with the authority to buy it.