Inside the ACE Developer Kit: a closer look at American Express’s approach to AI trust

Iulia Mușat, News Editor at The Paypers, delves into the most recent advancements in the AI world, covering American Express’s ACE Developer Kit and the broader implications of AI trust.

As more companies explore integrating AI into their systems, questions about liability, reliability, and oversight become unavoidable. Today, we’ll take a closer look at whether tools like the Agentic Commerce Experiences (ACE) Developer Kit are truly ‘a massive moment in commerce’ or merely a more structured way of managing risk.

From AI to agentic commerce

AI has become mainstream in recent years and is now used across various fields, including healthcare, automobiles, and marketing. It has also entered the finance space and taken root there,  with automated trading and transaction processing emerging as major developments in commerce and financial services.

During the second week of April, news broke that American Express took a step towards the ‘next era of commerce’. The company launched an agentic commerce developer kit and announced Amex Agent Purchase Protection, a commitment to expanding its support for Card Member purchases conducted by registered AI agents across its network.

Discussing with Luke Gebb, Executive Vice President and Head of Global Innovation at American Express, he noted that ‘American Express’s approach to agentic commerce combines clear Card Member intent with strong authentication, extending the same trusted protections customers rely on today into emerging AI-powered experiences. At the core of this approach is the concept of authenticated purchase intent: a structured, user-approved record that defines what an AI agent is allowed to do.’

However, questions remain about how liability is managed in specific cases, especially when distinguishing between user intent and execution error. According to American Express, if a Card Member authorises an AI agent to act on their behalf and the agent submits the customer’s authenticated purchase intent, the company can identify if a mistake originated from the agent.

In these particular cases, American Express states that it will safeguard eligible Card Members and merchants from charges resulting from AI agent errors. At the same time, merchants will be protected from chargebacks when an error is attributable to the agent operating within an authorised and authenticated transaction framework. 

In contrast, if an AI agent is compromised, the resulting transaction would be treated as fraudulent rather than an execution error.

Embedding trust and security in AI commerce

As detailed by American Express, the Amex ACE Developer Kit is set to offer select developers access to integrated solutions. These cover five main services, including Agent Registration, Account Enablement, Intent Intelligence, Payment Credentials, and Cart Context. 

Layer by layer, Agent Registration allows developers to verify AI agents so that only the trusted ones are authorised to transact on the Amex network. Account Enablement supports register cards for AI-driven transactions and facilitates more personalised experiences. On the other hand, Intent Intelligence helps ensure Card Member purchase intent is accurately captured to support authentication, approvals, and dispute handling.

Payment Credentials enable verified AI agents to make payments on behalf of cardholders using secure, tokenised data. Cart Context facilitates the sharing of cart details to optimise validation, authorisations, and dispute investigations.

American Express’s new offering circles around trust, except, as Gebb added, ’the company is not redefining the underlying liability model for digital commerce. Instead, it is layering new capabilities, such as agent verification, intent authentication, and enhanced transaction context, onto existing dispute and fraud frameworks.’

When AI gets it wrong: who’s accountable?

The question in everyone’s mind remains: can we trust AI with our payments, and who takes responsibility if an AI agent makes a mistake? With its new offering and a commitment to supporting customers in the agentic era, American Express promises to provide customers safeguards for registered agent purchases via Amex Agent Purchase Protection.

From the company’s perspective, ‘customers need confidence that they can define boundaries and be protected when things go wrong. This is why Amex Agent Purchase Protection is so important; it extends the same trusted protections customers rely on into agentic commerce.’

Circling back to our question, American Express claims that through Amex Agent Purchase Protection, the company will safeguard eligible customers from charges related to AI agent errors. However, at the time of writing, the terms and conditions of eligibility were not yet disclosed. The company also mentioned its plans to develop the foundation for agentic commerce to expand, while having transparency, accountability, and confidence at its core.

Minimising chargebacks: merchant protection mechanisms

Chargeback volumes have soared in recent years and will continue to do so, with Mastercard and Ethoca’s 2025 State of Chargebacks report estimating that, at a global level, disputes are set to increase from nearly 261 million in 2025 to 324 million by 2028. This 25% surge can be attributed to a variety of factors, including, but not limited to, the continuous rise in digital transactions, shifting consumer behaviour, friendly fraud and refund abuse, as well as the prevalence of card-not-present (CNP) transactions.

The entry of AI in commerce and the rise of AI agents have also changed the game for merchants. With AI agents now being able to conduct transactions independently, the traditional liability framework for chargebacks is put to the test. Merchants are required to confirm agent legitimacy and ensure that the transaction is authorised by the Card Member.

Retracing our steps to American Express, the company operates as an issuer, network, and acquirer under its closed-loop network. This allows it to handle the issues that arise from agentic commerce, supporting the provision of intent-driven authorisations, optimal fraud protection, and effective dispute resolution. In the end, this is set to result in enhanced safeguards for both Card Members and merchants.

Luke Gebb further explained that ‘for merchants, this approach is designed to reduce friction and risk over time. By introducing richer data signals, like intent and cart-level context, Amex aims to enable more accurate authorisations and fewer disputes.’

Agentic commerce: what comes next?

American Express has not been the only one pushing the narrative of agentic commerce. Throughout 2025 and until now, agentic commerce has increasingly positioned itself as the next frontier for companies across the financial sector.

As industry attention continues to rise, we asked American Express how its approach interoperates with emerging agentic standards, as well as how the company sees the technology evolving. Gebb told us that ‘from an ecosystem perspective, Amex is building for interoperability. The ACE Developer Kit is designed to align with and help shape emerging agentic standards, supporting integration across AI platforms, payment networks, and merchants.’ He added that ‘this reflects a recognition that agentic commerce will require shared protocols, consistent guardrails, and coordinated industry participation to scale effectively.’

From experimentation to real-world implementation, we are beginning to see new capabilities take shape, with the potential of bringing both advantages and risks to the table. A new era of commerce is emerging, where product discovery becomes more personalised, and the path from just interest to purchase requires fewer steps. Even if traditional interactions will not disappear overnight, how we find and buy products is evolving, maybe faster than we ever thought possible.

About Iulia Mușat

Iulia is a News Editor at The Paypers, covering fraud prevention, financial inclusion, and travel payments. Driven by a strong interest in financial security and payment innovation, she closely follows emerging trends, regulatory shifts, and technological developments shaping the industry. Through sharp analysis, Iulia brings insightful news and in-depth editorial pieces that help readers navigate and stay ahead in the rapidly evolving fintech space.