Major Banks Warn of Fraud Risks in AI-Powered Shopping Agents

Six major global banks have issued a cautionary report on AI shopping agents, warning that autonomous commerce systems carry significant risks around fraud, transparency, and consumer protection that intensify as these agents gain more independence.

The consortium, which includes Bank of America and Capital One, published "Building Trust in Agentic Commerce" on Tuesday, outlining concerns about AI agents that promise to fully automate online shopping. According to the report, consumers remain unclear whether these agents will act in their best interests, with particular worry that agents might purchase incorrect items, overspend, or expose users to scams.

The core concerns

The banks identify several structural problems with agentic commerce in its current form. AI agents could prioritize certain products or payment methods based on financial incentives like higher commissions or lower processing costs rather than customer benefit. The technology also creates potential for elevated rates of scams, fraud and transaction disputes, according to the report.

Data security represents another critical vulnerability. Any breach affecting these agents would compromise particularly sensitive consumer and merchant information, given the broad permissions and financial access required for autonomous shopping.

Recent security incidents underscore the risks

The timing of the report coincides with real-world validation of these concerns. Meta recently disclosed and patched a zero-day vulnerability in Muse, its AI assistant with shopping capabilities, that could have allowed attackers to hijack the system and exploit any user-granted permissions. The announcement came one day after Amazon asked Muse to stop accessing its e-commerce platform, citing concerns about how the agent handles customer credentials and fails to properly identify itself.

Ron Johnson, the former Apple executive who designed the original Apple Store concept, told TechCrunch he believes AI will improve online shopping information but won't fundamentally change consumer behavior. "AI will never be able to have you physically experience a product," Johnson said.

Why it matters

The banking industry's collective skepticism carries weight as financial institutions serve as gatekeepers for payment processing and fraud prevention. Their concerns signal that agentic commerce faces significant infrastructure and trust barriers before achieving mainstream adoption, despite aggressive development by AI companies and retailers. The gap between technological capability and consumer protection frameworks could slow deployment or trigger regulatory intervention.

Path forward requires standards

The banks aren't dismissing agentic commerce entirely. The report acknowledges the technology could potentially become "a mainstream way in which consumers and merchants transact." However, the consortium emphasizes that industry standards, policies and consumer protections must catch up to technological development first.

The six banks are now developing a follow-up paper detailing how to implement five key principles—transparency, safety, privacy and data protection, choice, and interoperability—to ensure safer deployment of AI-led e-commerce systems.

The details were first reported by Gizmodo.