The rise of agentic commerce is colliding with the rigid realities of financial risk management. A coalition of major banks including NatWest, Bank of America, Capital One, ING, ASB Bank, and Commonwealth Bank of Australia recently published a report outlining the severe vulnerabilities inherent in AI-driven shopping.
These institutions warn that autonomous agents are advancing far faster than existing industry standards and consumer protections can handle. The financial sector is drawing a hard line, signaling that the burden of securing these transactions will not fall solely on traditional banking networks.
Consumers Delegate Research But Reject Financial Autonomy
According to a recent report by the banking coalition, consumers are deeply unclear if AI will act in their interests.
Shoppers fear that agents might buy the wrong items, overspend, or lose money to scams. Research from PYMNTS Intelligence highlights this exact friction, noting that while 50 percent of Americans have made a retail purchase with AI assistance, only 24 percent are willing to let an AI agent handle both shopping and payment.
The consumer is comfortable delegating research, but much less comfortable delegating identity, payment choice, or an irreversible financial decision.
AI Agents Create New Vectors For Social Engineering And Fraud
The technical vulnerabilities are equally concerning. The banking coalition highlighted specific risks, such as AI agents requesting customer card details and entering them directly into third-party websites.
Bad actors could compromise or impersonate AI agents and merchants, deploying novel methods of social engineering to bypass traditional security checks. There is also a distinct risk that autonomous agents might steer users toward payment methods that carry fewer consumer safeguards, prioritizing transaction speed over security.
Liability Ambiguity Threatens The Entire Agentic Commerce Value Chain
When things go wrong, the allocation of liability remains unclear and inefficient. Dispute processes currently do not involve all relevant parties across the value chain.
If an AI agent makes an unauthorized purchase or falls victim to a scam, it is entirely ambiguous whether the merchant, the AI provider, or the issuing bank absorbs the loss. This ambiguity is a massive deterrent for financial institutions that are already grappling with rising fraud rates across digital channels.
Banks Are Building A Regulatory Moat To Protect Their Margins
The banking sector is not issuing these warnings purely out of altruism. By proposing new principles, such as mandatory disclosure when an AI agent is involved in a transaction and stricter data safeguards, these institutions are actively shaping the regulatory landscape.
They are building a defensive moat to ensure that the costly burden of agentic commerce fraud does not default to the traditional banking system.
Our Take
True Agentic Commerce Requires a Foundation of Trust that Current Technology has Not Yet Earned
The banking sector’s sudden concern over AI shopping agents is a predictable defense of its own profit margins. Financial institutions are happy to process the fees when AI drives seamless transactions, but they are rapidly retreating when the conversation turns to liability for autonomous fraud.
Retailers and tech companies must recognize that banks will not act as the shock absorbers for agentic commerce failures.
If you are building an AI shopping experience, you must design your own ironclad fraud prevention and liability frameworks, because the banks are already drafting the rules to ensure they do not pay for your mistakes.
