American Express Introduces Intent Intelligence Framework to Prevent Unauthorized AI Agent Purchases

American Express Tackles AI Shopping Risks with New Framework

American Express introduced the Intent Intelligence Framework on October 13, 2026, addressing a critical challenge as AI agents increasingly handle transactions on behalf of consumers1

1

. The framework aims to prevent AI agents from making unauthorized purchases by capturing and authenticating customer purchase intent before completing transactions2

2

. This proactive approach comes as the nondeterministic nature of large language models makes AI-driven shopping prone to errors and misinterpretations.

The company has not yet observed an increase in disputes from agentic commerce but wanted to prepare as this transaction model scales1

1

. Consumer interest signals readiness for AI-powered shopping: according to an Amex Trendex survey, 66% of surveyed US consumers said they're likely to use AI agents for price research, while 61% expressed interest in AI agents for product recommendations over the next 12 months1

1

.

How the Intent Intelligence Framework Operates

The Intent Intelligence Framework begins with the Create Intent API, which captures customer purchase instructions along with constraints like merchant name, maximum budget, and product preferences1

1

. Customers then authenticate their identity and approve the captured intent before the transaction proceeds2

2

.

Once verified, the system generates an Intent ID—a unique identifier representing the intent—and a Proof of Intent Token containing the customer's high-level constraints1

1

. This token obtains secure digital payment credentials that AI agents use at checkout instead of the customer's actual credit card number2

2

. The Intent ID and token support multiple functions across the transaction lifecycle, including fraud prevention and dispute resolution.

Preventing Errors Before Transactions Complete

Hannah Bozian, VP of Agentic Partnerships and Strategy at American Express, explained that the framework differs from other implementations by pulling intent earlier into the transaction flow, helping mitigate risk before it occurs1

1

. She offered an example: if a customer asks an agent to buy white Nike sneakers in size 9 but the agent attempts to purchase Adidas shoes, the system could prevent the transaction from completing because the agent made a mistake1

1

.

This capability becomes particularly valuable as AI agents move toward greater autonomy. Consumers might allow agents to reorder household staples like paper towels or toilet paper, where spending limits and product size matter more than specific merchants or brands1

1

. In these scenarios, having technology perform an extra verification check ensures transactions match customer intent without requiring manual approval for every purchase.

Building the Foundation for Agentic Commerce

The Intent Intelligence Framework is part of the broader Agentic Commerce Experiences (ACE Developer Toolkit), which American Express launched in April 20261

1

. The toolkit includes agentic tokens containing payment credentials and other services designed to support the payments ecosystem as it adapts to AI-powered transactions.

American Express designed the framework to remain flexible enough to support all agentic commerce models, regardless of which AI model is involved1

1

. The company is already in discussions with partners across the payments ecosystem for implementing ACE and plans to continue developing the platform as agentic commerce evolves.

Bozian emphasized that agentic transactions involve three parties—a consumer, a merchant, and an agent—and American Express wants to ensure these transactions remain seamless, trusted, and secure1

1

. The framework also maintains a log of what was authorized by which agent and when, which could help address fraud on behalf of the agent1

1

.

American Express hopes the Intent Intelligence Framework will become an industry standard for AI-powered transactions, setting expectations for how companies should handle customer intent verification as AI agents take on more purchasing responsibilities1

1

.