American Express releases tools to build AI payments—and pledges to pay the price if agents go awry | Fortune

Despite a surge of interest from technology hubs like Silicon Valley and financial powerhouses on Wall Street, agentic commerce—the paradigm where artificial intelligence agents autonomously execute purchases or other economic activities on behalf of a user—has thus far experienced relatively limited mainstream adoption. However, the profound potential for disruption within the online payments ecosystem is proving to be an opportunity too significant for many leading institutions to disregard. This burgeoning sector, characterized by its promise of unprecedented efficiency and automation, is now attracting substantial investment and strategic initiatives from established players, signaling a pivotal moment in its evolution.

Among the latest and most significant entrants is American Express, which on Tuesday unveiled a comprehensive agentic commerce developer kit designed to facilitate integration and foster innovation within this nascent field. Crucially, the financial services giant also announced a new purchase protection policy specifically tailored for erroneous transactions made by registered AI agents. This strategic move by American Express is not merely a product launch; it serves as a powerful indicator that what has largely been a niche industry is poised for rapid expansion and mainstream integration as the artificial intelligence era continues to unfold and mature.

Luke Gebb, Executive Vice President and Head of Global Innovation at American Express, articulated the current state of affairs to Fortune, remarking, "To date there have probably been as many press releases [on agentic commerce] as transactions, but no doubt it will happen. It will start within zones where there is just a good value for the user." Gebb’s candid assessment highlights the significant gap between industry excitement and actual consumer-facing deployment, yet simultaneously underscores an unwavering conviction in the technology’s eventual ubiquity, particularly in areas where it can deliver clear, tangible benefits to end-users.

The Emergence of Agentic Commerce: A Chronology of Industry Moves

The concept of agentic commerce, while seemingly futuristic, is rooted in the broader advancements of artificial intelligence and machine learning, particularly in natural language processing and decision-making algorithms. Its premise is simple yet transformative: instead of a human manually navigating websites, comparing prices, and completing payment forms, an AI agent, endowed with specific permissions and parameters, handles these tasks autonomously. This could range from ordering groceries based on inventory and dietary preferences, booking travel itineraries optimized for cost and convenience, to managing subscriptions and utility payments. The allure for consumers lies in reclaiming valuable time and reducing cognitive load, while businesses anticipate streamlined operations and enhanced customer engagement through highly personalized services.

The recent flurry of announcements from major financial players indeed corroborates Gebb’s observation about the proliferation of press releases. The timeline of these strategic moves paints a clear picture of an industry gearing up for significant change:

• April 2025: Mastercard made headlines with the unveiling of its "Agent Pay" initiative, a pioneering agentic payments technology specifically designed to power commerce in the age of AI. This early announcement positioned Mastercard as a frontrunner in developing the foundational infrastructure for AI-driven transactions, focusing on secure and seamless integration.
• Early 2026 (implied): Visa followed suit, announcing plans to "Open the Door to AI-Driven Shopping for Businesses Worldwide." Visa’s strategy emphasized enabling businesses to leverage AI for enhanced customer experiences and more efficient transaction processing, underscoring the B2B potential of agentic commerce alongside its consumer applications.
• March 2026: Payment processing giant Stripe entered the fray, releasing its own infrastructure protocols tailored for AI payments. Stripe’s involvement is particularly significant given its ubiquitous presence in online commerce, suggesting a push to embed agentic capabilities directly into the digital checkout flows that millions of businesses already use.
• This Week (Tuesday, implied 2026): American Express’s announcement of its developer kit and purchase protection policy marks the latest significant development. This move distinguishes Amex by not only providing foundational tools but also directly addressing one of the most pressing concerns for potential users: the financial risk associated with AI-driven errors.

This rapid succession of announcements from the world’s leading payment networks and financial institutions within a relatively short period indicates a collective belief that agentic commerce is not merely a fleeting trend but a fundamental shift in how economic activities will be conducted in the near future. It also suggests an intense competitive race to establish early dominance and set industry standards in this emerging domain.

The Promise and Peril of Autonomous Transactions

While the vision of AI agents seamlessly booking flights or automatically replenishing household staples offers considerable time-saving benefits for consumers, it concurrently introduces a complex layer of risks, particularly concerning instances where AI agents deviate from their intended tasks or make erroneous decisions. This inherent uncertainty surrounding AI agent behavior has, until now, been a significant hurdle for widespread adoption, and as Gebb acknowledged, the agentic commerce space has yet to fully resolve the critical question of liability and recourse for these errors.

This is precisely where American Express’s strategic intervention with its new purchase protection policy becomes a potential game-changer. By pledging to safeguard users from mistaken transactions initiated by AI agents registered on its platform, Amex directly confronts the core anxiety surrounding autonomous purchases. Gebb highlighted the company’s advantageous position in offering such protection, citing American Express’s long-standing and robust processes for handling disputed transactions from its card members. This existing infrastructure, honed over decades, provides a strong operational foundation for extending similar protections to AI-driven transactions.

When questioned about the potential financial cost to American Express of mitigating AI-related errors, Gebb expressed a forward-looking perspective. He posited that AI agents are expected to continuously improve in their accuracy and reliability over time. Furthermore, he argued that by instilling consumer confidence in the security and accountability of transacting through AI agents on the Amex network, the company stands to ultimately gain a substantial increase in overall transaction volume. This reflects a strategic investment in future growth, where the upfront cost of protection is offset by anticipated higher engagement and loyalty.

An American Express spokesperson further elaborated on the benefits of the new developer kit, asserting that it is designed to proactively reduce customer disputes and charge-backs. The kit’s architecture is intended to ensure that payment credentials are only provided to verified AI agents, and Amex will implement authentication protocols for card members before they are permitted to authorize agents to act on their behalf. This multi-layered verification process aims to provide merchants with a stronger baseline of legitimacy for agent-initiated transactions, potentially leading to fewer fraudulent activities and contested charges.

Supporting Data and Market Projections

The nascent stage of agentic commerce makes precise market sizing challenging, but broader trends in AI adoption and e-commerce offer compelling indicators of its potential trajectory. According to various industry analyses, the global AI market is projected to reach hundreds of billions of dollars within the next few years, with significant growth in applications related to automation, personalization, and customer service. E-commerce itself continues its relentless expansion, driven by convenience and digital innovation. Agentic commerce stands at the intersection of these two mega-trends, poised to capture a substantial share of future digital transactions.

Analysts suggest that early adoption will likely occur in areas with high-frequency, low-value transactions, or those requiring complex comparison and scheduling. Examples include:

• Subscription Management: AI agents could automatically renew or cancel subscriptions based on usage patterns, price changes, or user preferences.
• Inventory Management (Home/Small Business): Automated reordering of consumables when supplies run low, optimizing for cost and delivery time.
• Travel Planning: AI agents could monitor flight and hotel prices, book optimal itineraries, and manage changes based on user calendars and preferences.
• Personalized Shopping: AI agents could proactively identify and purchase items tailored to a user’s style, needs, or past purchases, often discovering better deals or unique products.

However, widespread consumer adoption hinges on overcoming several significant barriers:

1. Trust and Transparency: Users need to understand how AI agents make decisions and feel confident that their financial interests are protected.
2. Security and Privacy: Robust measures are essential to prevent unauthorized access, data breaches, and misuse of financial information.
3. Interoperability: A fragmented ecosystem of agents, platforms, and payment networks could hinder seamless integration.
4. Regulatory Clarity: Governments and financial authorities will need to establish clear guidelines for AI liability, consumer rights, and data governance in this new commerce model.

The strategic investments by companies like Amex, Mastercard, Visa, and Stripe are not merely speculative bets; they are calculated moves to address these barriers and lay the groundwork for a trusted and efficient agentic commerce ecosystem. Their involvement signals a serious commitment to standardizing protocols, enhancing security, and building the necessary infrastructure to scale this technology.

Broader Impact and Future Implications

The advent of agentic commerce, propelled by the entry of major financial institutions, promises to reshape the landscape of digital transactions and consumer behavior in profound ways.

• Transformation of Online Shopping: The passive act of browsing could evolve into an active, intelligent delegation. Consumers might shift from searching for products to defining needs that AI agents then fulfill, leading to a highly personalized and automated shopping experience. This could favor merchants who integrate seamlessly with agentic platforms and offer competitive value.
• Competitive Dynamics: Traditional e-commerce platforms, payment processors, and even individual retailers will need to adapt rapidly. Companies that can provide APIs and data feeds that AI agents can easily interpret and interact with will gain a significant advantage. The race will be on to become the preferred network for AI-driven transactions. Fintech innovators, often at the forefront of digital payments, will face new competition but also new opportunities to develop specialized AI agents or infrastructure components.
• Regulatory Frameworks: The current legal and regulatory environment is largely designed for human-initiated transactions. Agentic commerce introduces novel questions about liability (who is responsible when an AI makes an error?), consumer protection (how are users informed and protected from autonomous decisions?), and data governance (how is data collected and used by agents?). Governments worldwide will likely need to develop new frameworks to address these complexities, balancing innovation with safety and fairness.
• Ethical Considerations: Beyond legalities, ethical questions will arise. Could AI agents inadvertently perpetuate biases present in their training data, leading to discriminatory purchasing patterns? What are the implications for consumer autonomy if increasingly significant decisions are outsourced to algorithms? These considerations will require ongoing dialogue and proactive design choices to ensure responsible development.
• The Role of Stablecoins: Interestingly, some voices within the crypto community advocate for the agent economy to operate on stablecoin rails. Stablecoins, cryptocurrencies pegged to stable assets like the U.S. dollar, offer the promise of instant, low-cost, and borderless transactions. Gebb acknowledged American Express’s "variety of plans" surrounding stablecoins, though he views them as more suitable for settlement processes rather than direct consumer payments, declining to provide specific details. Crypto proponents argue that stablecoins could facilitate efficient micropayments essential for an agent economy, where numerous small, automated transactions might occur between different AI agents or services. Their programmatic nature and global accessibility could offer a compelling alternative to traditional payment networks for certain types of agentic commerce, though regulatory hurdles and scalability challenges remain significant.

In conclusion, American Express’s strategic foray into agentic commerce, characterized by both enabling technology and robust consumer protection, signifies a critical juncture for the industry. While the journey from "press releases to transactions" is still unfolding, the commitment from major financial players suggests that AI-driven autonomous commerce is no longer a distant dream but an impending reality, poised to redefine how we interact with the digital economy and manage our financial lives. The coming years will undoubtedly witness a rapid acceleration in development, adoption, and, crucially, the establishment of the necessary trust and regulatory guardrails to support this transformative shift.