American Express launches AI purchase authorization framework for agentic commerce

American Express launches AI purchase authorization framework for agentic commerce
Amex unveils AI intent checks

As AI shopping tools begin to handle more transactions on behalf of consumers, payment providers are moving to reduce the risk of unauthorized purchases. American Express says its new Intent Intelligence Framework is designed to verify a customer’s purchase intent before an AI agent completes a transaction and could also support fraud prevention and dispute resolution.

Highlights

  • American Express introduced the Intent Intelligence Framework, leveraging a Create Intent API to secure AI-driven purchases by capturing customer constraints and using digital credentials instead of card numbers.
  • The framework integrates into the ACE Developer Toolkit and addresses potential agentic commerce disputes by verifying intent earlier in the payment flow to block unauthorized transactions.
  • Amex Trendex survey shows 66% of U.S. consumers likely to use AI agents to research prices and 61% for recommendations in the next 12 months, supporting the upcoming demand for the framework.

Framework design and transaction flow

As first reported by TechCrunch, American Express is introducing the framework as part of a broader push to make AI-driven commerce secure as autonomous shopping tools gain traction. The company says the system is meant to preserve a customer’s instructions throughout the full transaction lifecycle, from the initial request to checkout and any follow-up issues.

Called the Intent Intelligence Framework, the model begins with a Create Intent API that captures purchase instructions and any constraints attached to the request, including merchant name, spending limit and product preferences. After that, the customer authenticates their identity and approves the intent gathered by the AI agent.

American Express then creates an Intent ID and a Proof of Intent Token, which reflects the customer’s high-level constraints. That token is used to obtain a secure digital payment credential for checkout instead of the customer’s actual card number, with the aim of keeping the transaction aligned with what the user authorized.

Hannah Bozian, vice president of agentic partnerships and strategy at American Express, says the company’s approach brings intent verification earlier into the payment flow. She says that could stop transactions before completion if an agent misinterprets a request, such as attempting to buy a different brand than the one the customer specified.

Industry adoption and payments impact

American Express says it has not yet seen a rise in disputes tied to agentic commerce, but it is building the technology ahead of broader adoption. The framework follows the company’s April launch of its Agentic Commerce Experiences, or ACE Developer Toolkit, which now includes the intent capabilities.

The company says consumer interest in AI-assisted shopping is already visible in its recent Amex Trendex report. In that survey, 66% of U.S. consumers say they are likely to use AI agents to research the best price for a product or service, while 61% say they are likely to use them for recommendations over the next 12 months.

American Express argues the framework could become more important as AI agents act with greater autonomy in routine purchases, such as household essentials where price or size may matter more than brand or merchant. The company also says a record of what was authorized, by which agent and when, could strengthen fraud controls and streamline dispute handling across the payments ecosystem.

American Express says it is already discussing implementation of ACE with partners across the payments industry and plans to continue developing the toolkit as agentic commerce evolves.

In our earlier article on scientists calling for an international body to oversee containment standards for advanced cyber-capable AI, we explained why researchers argue voluntary safety pledges are not enough. The piece highlighted proposals for enforceable containment requirements, independent oversight, and incident reporting to reduce the risk of AI systems bypassing safeguards as capabilities advance.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.