AI agents crash the payments party—but who picks up the tab?

Agentic commerce fundamentally reshapes traditional payment liability by introducing AI agents as a 'fifth participant' that interposes between cardholders and merchants, thereby shifting risk downstream to merchants who remain the Merchant of Record despite losing direct control over transaction initiation and verification. While protocols like OpenAI and Stripe’s Agent-to-Commerce Protocol (ACP) deploy scoped, ephemeral Shared Payment Tokens to safeguard consumer credentials, they fall short in fully mitigating merchant exposure to intent disputes, underscoring a critical gap in protecting merchants within autonomous payment flows.

The disappearance of human behavioral signals such as typing cadence and mouse movements in agentic commerce erodes traditional fraud detection mechanisms, leading to a complete loss of trusted behavioral biometrics and device fingerprinting. This absence heightens vulnerability to 'friendly fraud' and hallucination disputes, compelling the payments industry to urgently innovate new identity verification frameworks that can authenticate both the AI agent's legitimacy and the consumer's real-time intent, as emphasized by Visa’s Michele Herron who stresses the need for infrastructure that unifies identity, risk signals, and contextual data.

By mid-2026, major players including Alipay, Visa, Highnote, Mastercard, and Stripe have launched pioneering protocols and platforms that weave together cryptographic innovations, tokenization, and blockchain technology to establish secure, scalable, and standardized payment ecosystems for autonomous AI agents. Alipay’s Agentic Commerce Trust Protocol and Token Pay exemplify AI-native payment methods enabling user-governed transactions, while Visa and Highnote’s programmable payment controls and Mastercard’s Agent Pay protocol leverage Polygon blockchain and partnerships with Coinbase and Stripe to support diverse payment methods including stablecoins, cards, and bank accounts, collectively building a layered capability ladder that accommodates varying merchant risk tolerances.

Stripe’s Agent-to-Commerce Protocol (ACP) represents a critical technical foundation by replacing unreliable UI scraping with a standardized API-driven communication layer that expresses product catalogs in structured JSON, allowing AI agents to programmatically select products, manage checkout states, and execute payments via flexible methods such as crypto and cards. This approach preserves merchant control over customer relationships and risk signals, facilitating safe autonomous transactions and minimizing disputes, while broader industry momentum—spanning Stripe, Visa, Mastercard, and Coinbase—signals a concerted push toward interoperable, cryptographically secured payment infrastructures that underpin the future of agentic commerce.

By early 2026, major payment networks Visa and Mastercard spearheaded pilot programs and strategic partnerships to embed AI-driven agentic payment models within banking and commerce ecosystems. Mastercard’s 'Agent Pay for Machines' (AP4M) initiative, launched with over 30 partners including Coinbase, Polygon, and Stripe, exemplifies a coordinated industry effort to enable autonomous AI agents to execute micropayments using blockchain technology, supporting diverse payment methods such as cards, bank accounts, and stablecoins. Concurrently, Visa’s collaboration with FPT AI Factory and InFlow to create an agent-native commerce platform highlights practical deployments where AI agents autonomously research, procure, and pay for services, ensuring secure and auditable transactions. These initiatives collectively signal a fintech pivot towards operationalizing AI-mediated payments at scale, blending traditional payment rails with emerging blockchain and AI capabilities.

Complementing these network-led efforts, fintech innovators and platform providers are advancing ecosystem readiness through specialized tools and protocols that facilitate agentic commerce adoption. Stellagent’s Agentic Commerce Studio, targeting merchants and payment service providers in Asia, focuses on validating emerging AI commerce protocols such as ACP, UCP, AP2, Visa TAP, and Mastercard Agent Pay, thereby fostering interoperability and readiness across the payment ecosystem. Meanwhile, Clink’s unified payment and billing platform empowers AI builders to accept payments globally from both humans and authorized AI agents, supporting over 135 currencies and 100 local payment methods. Its integration with coding agents like Cursor and Claude Code automates payment integration, dramatically lowering engineering barriers and accelerating practical deployments without requiring merchants to abandon existing processors.

In the Asian market, Alipay’s launch of the world’s first AI Wallet, Token Pay, and the Agentic Commerce Trust Protocol marks a significant milestone in establishing standardized frameworks for AI-driven transactions. Developed in partnership with AI firms MiniMax and Stepfun, Alipay’s ecosystem combines intelligent security systems with a global subscription payment platform tailored for AI companies, facilitating token top-ups and seamless agentic commerce at scale. This initiative underscores the importance of cross-industry collaboration in creating trust and operational infrastructure necessary for widespread agentic commerce adoption, positioning Alipay as a pioneer in integrating AI agents into mainstream payment flows.

Beyond the major networks and platforms, targeted fintech partnerships such as b1BANK’s collaboration with Covecta demonstrate how agentic AI is being deployed to enhance automation, decision support, and customer engagement within banking operations. These practical implementations reflect a broader industry trend where AI agents are not only transforming payment mechanics but also enriching customer experiences and operational efficiencies, thereby laying the groundwork for agentic commerce to move from experimental pilots to everyday financial interactions.

By mid-2026, AI-driven data orchestration has become foundational to agentic commerce platforms, as exemplified by the partnership between FPT AI Factory, InFlow, and Visa Intelligent Commerce which launched an agent-native commerce platform enabling autonomous end-to-end procurement with secure, auditable payments and dynamic AI model routing. This evolution transforms issuer processing from a back-end utility into a real-time decisioning layer that leverages behavioral, contextual, and credential data to optimize payment authorization, reduce fraud, and personalize commerce experiences without human intervention.

Despite the promise of AI orchestration, nearly half of organizations (47%) struggle with poor-quality data, which hampers effective decision-making and increases friction in smart checkout processes. To address this, agentic commerce systems now incorporate real-time identity verification, spending controls, and contextual signals to validate AI-initiated transactions autonomously, aiming to reduce false declines that currently cost the industry an estimated $30 billion in lost sales.

Consumer research underscores the transformative impact of AI-driven smart embedded offers on shopping behavior, with 81% of consumers indicating such offers influence their choice of where to shop and 77% saying real-time savings would affect their default payment method. However, a significant visibility gap persists as only 10-17% of shoppers notice payment-linked offers, contributing to $42.4 billion in missed or unredeemed discounts annually. Preferences also lean toward low-risk AI automation that applies the best payment method or rewards rather than substituting products, highlighting the need for personalization guardrails focused on payment and offer application.

Issuers are highly enthusiastic about smart checkout innovations, with 90% expressing strong interest and 87% expecting these technologies to become core to their rewards strategies within five years. Smart checkout transforms rewards from post-purchase perks into real-time incentives at the payment decision point by integrating transaction, payment, and loyalty data to deliver personalized offers that increase basket size and influence payment choice. However, issuers emphasize the necessity of clear governance, transparency, and controls over data use and financial exposure to build trust, anticipating that increased brand- and retailer-funded incentives will reshape rewards economics and drive greater investments in technology and oversight.

The checkout experience is rapidly evolving into a critical AI infrastructure that prioritizes frictionless, seamless payment authorization through machine-readable pricing, interoperable APIs, and embedded loyalty mechanisms. With 84% of global shoppers valuing one-click checkout and 94% interested in smoother payment innovations, merchants and AI agents alike are focusing on reducing authentication barriers and integrating loyalty directly into transactions to minimize abandonment and boost repeat engagement. In this AI-mediated economy, transactional efficiency and fulfillment accuracy eclipse traditional brand loyalty, positioning smart checkout as the new front door for consumer-agent interactions.

Agentic commerce disrupts the traditional four-party payment model by introducing AI agents as a 'fifth participant,' fundamentally shifting liability downstream to merchants who are left accountable when autonomous agents make erroneous purchases without explicit user consent. As highlighted in the March 2026 analysis, "liability follows the Merchant of Record," placing merchants in a precarious position where they bear the financial risk despite lacking direct control over the agent's actions.

Traditional fraud detection techniques, reliant on behavioral biometrics like typing cadence or device fingerprinting, become obsolete in agentic commerce since transactions are initiated by software agents rather than humans. This erosion of conventional risk signals creates a critical vulnerability for merchants, who must now contend with a 'shifting trust perimeter' as payments originate outside their controlled storefront environments, forcing them to trust opaque AI agents they cannot inspect, thereby exposing structural governance and compliance failures.

Emerging technical standards such as OpenAI and Stripe’s Agent Commerce Protocol (ACP) offer promising frameworks to standardize communication between AI agents and merchants, primarily by protecting consumer credentials through mechanisms like the Shared Payment Token (SPT). However, as the March 2026 report notes, these protocols currently fall short of shielding merchants from disputes over intent and liability, underscoring the need for more comprehensive solutions that address merchant risk in agentic transactions.

By mid-2026, the Center for Financial Enterprise Security (CFES) advanced the governance conversation by applying common-law principal-agency principles to clarify oversight and accountability in agentic commerce, advocating for verification of agency relationships and defined scopes of authorization. While existing payment laws provide a foundational framework, CFES emphasizes the necessity of industry-led collaborative efforts, including ongoing dialogues and webinars facilitated by FS Vector, to fill regulatory gaps, manage fraud risks, and build operational resilience and trust in this evolving payment ecosystem.

By early 2026, payment leaders recognized that embedding value-added services beyond basic money movement was essential to thrive in the agentic commerce era. This included monetizing data products and seamlessly sharing them across ecosystems, as well as solving complex identity and trust challenges to distinguish between human and AI agents at scale—a capability Visa emphasized as critical to confirming agent legitimacy and consumer intent. Banks and issuers faced pressure to build AI-compatible agents or risk losing transaction routing to more prepared competitors, highlighting a strategic pivot toward integrating AI into core payment infrastructures.

Issuers pulling ahead in this new landscape are embedding their payment products directly into customers’ daily money flows—linking cards to payroll, subscriptions, and gig income—to create more durable relationships and increase lifetime value. High-performing issuers focus on deepening existing relationships through AI-driven personalization and embedded onboarding rather than solely chasing new customer acquisition, with 82% pursuing both strategies simultaneously. AI-powered real-time transaction categorization and enrichment have become the most widely adopted capabilities, underscoring the shift from rewards and promotions to securing 'default status' as the financial product embedded in everyday spending.

The rise of AI-driven commerce is reshaping rewards and checkout experiences, with issuers aggressively adopting smart checkout technologies that embed personalized rewards and incentives directly into the payment flow. Ninety percent of issuers express strong interest in smart checkout, which transforms the checkout moment into a dynamic, AI-powered recommendation engine optimizing incentives in real time based on transaction data, loyalty, credit risk, and contextual factors. This evolution allows issuers to reclaim strategic relevance by controlling loyalty economics more effectively than traditional advertising, as three-quarters expect smart basket systems to increase their influence over consumer payment choice and merchant selection.

Fintech innovators and payment networks like Stripe, Visa, Mastercard, and Coinbase are actively exploring new payment methods, including stablecoins and cryptographic primitives, to enable AI agents to transact autonomously and operate 24/7 beyond traditional payment rails. This technological shift is accompanied by a growing emphasis on embedding provenance, identity, privacy, and reputation services into payment ecosystems to meet the unique demands of agentic commerce. The recent milestone of Base reporting 3.1 million AI agent payments in 30 days exemplifies how autonomous payment capabilities are becoming integral to evolving business models targeting new AI-driven customer segments.

By mid-2026, the trajectory of agentic commerce is clearly moving toward a landscape where machine-initiated payments dominate, leveraging blockchain-based rails to enable nano payments and dynamic agent marketplaces. However, the complexity of integrating these on-chain systems with traditional payment rails such as credit cards and bank accounts introduces significant regulatory challenges, as the rules governing such hybrid transactions remain unclear. This regulatory ambiguity underscores the need for clearer frameworks to support the seamless operation of agentic commerce across diverse payment infrastructures.

The future winners in agentic commerce will be those platforms that master the seamless integration of blockchain and traditional payment rails, enabling AI agents to fluidly optimize payment flows across stablecoins, credit cards, and bank details. B.AI exemplifies this trend by pioneering frictionless global Web3 payments that bridge decentralized and conventional systems, demonstrating a practical pathway to scaling agentic commerce on a global scale. This dual-rail integration not only enhances transactional efficiency but also positions companies to capture value from both the underlying commodities and the payment rails themselves.