American Express is an American financial services company headquartered in New York City, New York. The company has more than 76,800 employees worldwide and reported sales of $18.98 billion in the fourth quarter of 2025, with a sales growth outlook of 9-10% for fiscal year 2026.
American Express began applying machine learning to fraud detection in 2010, making it one of the earliest adopters of AI among financial services companies. Today, the company has provided nearly all of its 76,800 employees worldwide with access to leading AI tools, expanded its AI-assisted development tools to more than 11,000 engineering professionals, and reduced coding cycle times by more than 30%.
Machine learning-powered fraud detection models monitor over $1.2 trillion in transaction value each year and generate fraud verdicts for every card transaction worldwide in milliseconds. The company is currently exploring more than 70 generative AI use cases across its organization and, through Amex Ventures, is investing in generative AI startups focused on trust and safety, enterprise efficiency, and data-driven experiences.
For financial institutions operating at a global transaction scale, AI systems are increasingly impacting how they operationally manage fraud risk, transaction approvals, and customer retention workflows. For American Express, many of these systems are integrated directly into mass decision-making processes, rather than being positioned as standalone AI initiatives.
Publicly available reports and company documents point to two particularly notable applications of AI within organizations.
- Real-time fraud detection during transaction authentication
- Predictive customer retention and seller targeting
In both cases, machine learning systems examine large amounts of behavioral and transactional data to support faster operational decision-making, reduce manual review workloads, and improve customer experience outcomes.
Real-time fraud detection during transaction authentication
Fraud prevention is one of the hottest AI applications within American Express. The company has been looking at using machine learning systems to evaluate transactions in real-time during payment approval workflows.
According to an analysis by the Harvard Business School Digital Initiative, American Express applies machine learning models to analyze purchasing patterns, spending behavior, merchant activity, and transaction anomalies during approval requests.
The business issues are critical for payment providers. Fraudulent transactions cause direct financial loss, while incorrectly rejecting legitimate purchases can erode customer trust and transaction volume.
AI systems likely process multiple forms of transactional and behavioral data, including:
- Past transaction records
- Frequency and speed of spending
- Seller category activity
- Geographic purchasing patterns
- Device and account information
- Real-time authentication signal
Rather than relying solely on static fraud rules, machine learning models assess whether transactions deviate from expected customer behavior.
An NVIDIA case study of American Express’s infrastructure shows that the company uses a GPU-accelerated AI system that can process fraud decisions within milliseconds.
For customers, the impact on workflow is largely invisible until suspicious behavior is identified. Legitimate purchases are processed through approval, while high-risk transactions may trigger additional verification or review.
For anti-fraud teams, machine learning systems can help prioritize transactions that require human investigation. This reduces the amount of low-risk transactions that require manual review and allows analysts to focus on more complex fraud cases.
The use of AI in fraud detection is consistently discussed in internal documentation, academic analysis, and infrastructure partner reports, suggesting that machine learning plays an operational role in American Express’ transaction security processes.
Reduce manual review workload – Machine learning systems can help fraud teams focus investigative resources on higher-risk transactions rather than reviewing large volumes of routine work.
Improving transaction approval accuracy – Behavioral models help distinguish between legitimate purchase anomalies and potential fraud during approval workflows.
Agent commerce infrastructure for AI-powered trading
American Express is also investing in what it calls “agent commerce.” This is a model in which AI agents can perform commerce-related tasks on behalf of customers, such as purchasing goods, booking travel, making reservations, and making payment transactions. Rather than focusing on customer analytics, this effort is focused on creating the infrastructure necessary for AI agents to safely participate in commerce.
This business problem stems from the challenges facing the emerging agent AI ecosystem. Traditional payment systems are designed around direct human action, where customers explicitly select a product, enter payment details, and approve the purchase. Once AI agents can perform tasks on behalf of users, payment providers must establish mechanisms to verify the agent’s identity, authenticate customer intent, and maintain transaction security.
To address this challenge, American Express introduced the Agentic Commerce Experiences (ACE) developer kit. The platform includes five interconnected features designed to support agent-driven transactions, according to a company statement.
- Agent registration and verification
- Activate customer account
- Intent validation and authentication
- Issuing tokenized payment credentials
- Transaction context and authorization control
The data processed within these workflows is different from that of traditional fraud detection systems. In addition to transaction and payment information, the platform evaluates the customer’s purchase intent, approval authority, agent credentials, and transaction context before allowing the AI agent to complete the purchase.
For customers, the intended workflow will move from performing all transactions manually to defining purchase intent and approval parameters that can be acted upon by a validated AI agent. American Express says customers will be able to manage spend management, purchase approvals, and active AI agent authorization through digital channels.
