American Express framework records customer approval for AI-led purchases
American Express has introduced its Intent Intelligence Framework to record and authenticate a customer’s instructions before an AI agent makes a purchase. The company wants the framework to help keep agent-led transactions within the limits customers set as AI assistants increasingly shop on their behalf.
The process captures instructions such as the merchant, budget and product preferences, then asks the customer to verify their identity and approve the recorded intent. Amex creates a unique Intent ID and a Proof of Intent Token, which can be used to obtain a secure digital payment credential instead of sharing the card number. The company says it has not seen more disputes from agent-led shopping so far, but is preparing for its growth; its survey found 66% of US consumers were likely to use agents to research prices and 61% for recommendations in the next year.
- Amex records and verifies purchase instructions before an AI agent checks out.
- Approved limits can include the merchant, budget and product preferences.
- 66% of surveyed US consumers may use AI agents to compare prices.
New here? Start with this
Artificial intelligence agents that can shop on behalf of customers are becoming increasingly common. These AI assistants research products, compare prices and complete purchases with minimal human involvement at each transaction. As this practice grows, keeping these transactions within customers' wishes whilst maintaining security is becoming important.
American Express has created a system to verify customer approval before an AI agent makes a purchase. The system records the customer's preferences, including the merchant, budget and product type, then asks them to verify their identity and confirm approval. Rather than sharing card details with the AI agent, American Express instead issues a secure digital token for use in that specific transaction.
This development reflects expectations of rapid growth in AI-assisted shopping. Recent surveys found that roughly two-thirds of American consumers plan to use AI agents to research and compare prices in the coming year, with many also seeking product recommendations from these systems. Although the company has not yet observed a significant rise in disputes from such purchases, the system is designed to prevent problems before they become widespread.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
The Intent Intelligence Framework represents a responsible approach to authenticating AI-driven commerce. By requiring customers to explicitly verify their instructions before authorisation, alongside setting specific budgets and preferences, the system maintains genuine consumer agency rather than handing over purchasing authority wholesale to algorithms. The use of Intent ID tokens and Proof of Intent rather than sharing card numbers reduces fraud risk and protects customer financial data, whilst the recorded intent creates clear accountability if disputes arise. American Express is prudently preparing for the inevitable growth of agent-led shopping by implementing guardrails now, preventing a future landscape where customers routinely discover unauthorised or undesired purchases.
The case against
Whilst presented as consumer protection, the framework may create a false sense of security that obscures the actual risks of delegating purchases to AI systems. Recording and storing detailed intent data concentrates sensitive information about customer preferences and spending patterns with financial institutions, creating new privacy vulnerabilities and opportunities for algorithmic profiling. The approval mechanism itself is only as strong as the customer's genuine review of it—in practice, users typically rubber-stamp authorisations without careful scrutiny, rendering the framework largely symbolic. More fundamentally, the framework doesn't address whether AI agents should be making autonomous purchasing decisions at all, merely adding procedural layers that may make problematic technology seem legitimate without preventing the underlying harm to consumer autonomy and the shift of purchasing power away from individuals towards corporations controlling the AI systems.
Read the full article at the source →
Originally published by TechCrunch as “Amex introduces a framework to prevent AI agents from making unauthorized purchases”.