Global Banks Warn AI Shopping Bots Increase Fraud Risks

A consortium of six global banks, including Bank of America, Capital One, ING Group, and NatWest Group, published a principles paper titled "Building Trust in Agentic Commerce" on September 22, 2026. The report warns that AI agents designed to automate online shopping introduce significant risks regarding transparency, safety, privacy, and interoperability.

The banks highlight a growing gap between the rapid advancement of agentic commerce and the development of industry standards. Specific concerns include AI agents requesting credit card details directly, steering users toward payment methods with weaker protections, and the potential for increased credit card disputes. The consortium notes that consumers may be unsure if these agents will act in their interests or how liability is handled when an agent exceeds its authority.

These warnings follow security incidents involving Meta's AI assistant, Muse, which required a patch for a zero-day vulnerability that could have allowed attackers to hijack the tool. Additionally, Amazon blocked Muse from its platform after the agent failed to identify itself during shopping activities. To mitigate these risks, the banks propose mandatory disclosures when AI agents are involved in transactions, increased transparency in AI decision-making, and enhanced data safeguards. Despite these warnings, the banks expressed a willingness to work with stakeholders to establish the protections necessary for AI shopping to become a mainstream transaction method.