📊 Key Data
  • 31% higher fraud declines: Fraud flagged transactions surged 31% during the 2025 Cyber Five holiday weekend compared to normal days.
  • 52% fear fraud: Over half of consumers refuse AI shopping due to fraud concerns.
  • $15 trillion B2B spend: Gartner projects 90% of B2B buying will be intermediated by AI agents by 2028.
🎯 Expert Consensus

Experts agree that while AI-powered commerce presents significant economic opportunities, its success hinges on robust fraud prevention, clear liability frameworks, and consumer trust.

4 days ago
The Ghost in the Checkout Line: Solving the Identity Crisis of AI Commerce

The Ghost in the Checkout Line: Solving the Identity Crisis of AI Commerce

DURHAM, NC — November 17, 2026 — For the past two decades, the fundamental assumption of digital commerce has been that a human is sitting on the other side of the screen. Retailers optimized their checkout flows, fraud models, and legal frameworks around the predictable, if sometimes erratic, behavior of people. But as artificial intelligence evolves from a conversational assistant into an autonomous actor, a new kind of consumer is arriving at the digital storefront: the AI shopping agent.

This shift, broadly termed "agentic commerce," promises to eliminate the friction of routine purchasing. Yet, it also introduces a profound identity crisis for merchants. When a headless browser or an API call attempts to buy a pair of shoes or book a flight, how does the retailer know if it is a legitimate AI assistant acting on behalf of a loyal customer, or a malicious botnet attempting to validate stolen credit card credentials?

This week, a new partnership between payment orchestration platform Spreedly and fraud intelligence firm Riskified offered an early blueprint for how the industry plans to solve this dilemma. By integrating Riskified's AI Agent Intelligence directly into Spreedly's Composer platform, the companies are establishing dedicated verification rails for non-human purchasers. It is a necessary infrastructure upgrade for an industry that is quickly realizing that AI-powered commerce cannot scale on novelty alone; it requires a bedrock of trust and specialized risk management.

The Technical Dilemma of the Machine Shopper

The line between a helpful AI agent and a harmful bot is technically razor-thin. Both rely on automated scripts, both can execute tasks at superhuman speeds, and both often obscure their IP origins. For years, fraud prevention platforms have trained their machine learning models to identify and block exactly this type of non-human behavior.

"AI shopping will only succeed if consumers feel confident handing over their purchasing decisions," said Justin Benson, CEO of Spreedly. "That means making it easier for merchants to recognize legitimate transactions, stop fraud and protect their customers, without slowing down the consumer experience."

The pressure on merchants is already mounting. According to anonymized transaction data from Spreedly, declines flagged for suspected fraud averaged 31% higher per day over the 2025 Cyber Five holiday weekend compared to a normal November day, even as overall payment attempts rose by 20%. Retailers are caught in a crossfire: their existing security systems are likely to block legitimate AI agents as fraudulent bots, resulting in lost sales and frustrated early adopters.

The Spreedly and Riskified integration attempts to thread this needle. By placing fraud decisioning as a distinct step before any payment is processed, merchants can utilize a policy builder specifically designed for agent-driven orders. They can choose to automatically approve known, low-risk agents, decline suspicious automated traffic, or "step up" the authentication process—perhaps by pinging the human user's smartphone for a biometric approval before the AI is allowed to finalize the transaction.

When Algorithms Buy: The Liability Vacuum

While the technical hurdles of identifying AI agents are actively being solved, the legal and financial frameworks surrounding them remain dangerously ambiguous. If an AI agent hallucinates and purchases a thousand dollars worth of incorrect merchandise, who pays the bill?

Consumers are acutely aware of this gray area. According to Spreedly's State of Checkout 2026 survey, conducted by Talker Research among 2,000 U.S. adults, 52% of respondents cited the fear of fraud as the primary reason they refuse to let AI shop for them. More tellingly, 31% explicitly stated they do not know who is responsible when an AI purchase goes wrong. Independent research from Riskified mirrors this hesitation, showing that only 16% of shoppers are currently comfortable with an AI agent spending their money autonomously.

"The concept of digital agency is going to severely test existing consumer protection laws," noted one commercial legal expert specializing in fintech. "Credit card networks like Visa and Mastercard have robust dispute resolution rules for 'unauthorized' transactions. But if a consumer gives an AI broad permission to manage their grocery shopping, and the AI makes a bizarre, expensive error, is that transaction truly unauthorized? The human authorized the agent, even if they didn't authorize the specific outcome."

Currently, the Federal Trade Commission and major card brands have not issued definitive, updated guidelines specifically addressing the nuances of autonomous AI purchases. This regulatory vacuum means that early disputes will likely be messy, falling back on traditional chargeback codes like "not as described" or "unauthorized transaction." For merchants, this ambiguity is a massive financial risk. If consumers blindly dispute AI-driven errors, merchants could face crippling chargeback fees and potential penalties from payment processors.

Court or Block: The Merchant's Strategy Tightrope

Despite the risks, the economic incentive to welcome AI agents is simply too massive to ignore. While consumer willingness for AI to make direct, high-stakes purchase decisions remains relatively low in the B2C sector, the B2B landscape is accelerating rapidly. Gartner projects that by 2028, 90% of B2B buying will be intermediated by AI agents, representing over $15 trillion in spend. Furthermore, analysts forecast that up to 20% of all digital commerce transactions will run through autonomous platforms by the end of the decade.

"Merchants shouldn't have to choose between welcoming AI shoppers and protecting their business," said Max Meister Admoni, Global Head of Partnerships at Riskified. "By pairing Riskified's risk intelligence with Spreedly's payment orchestration, merchants get the visibility and control to set their own rules for agent-driven orders."

This level of control is paramount. A luxury retailer might choose to entirely block autonomous agents to preserve exclusivity and human touch, while a wholesale distributor might want to fast-track agent orders to maximize volume. Payment orchestration platforms—which sit above individual payment gateways and allow merchants to route transactions dynamically—are uniquely positioned to enforce these complex, conditional rules.

Building the Infrastructure for a Post-Human Web

The Spreedly and Riskified partnership does not exist in a vacuum. It is part of a broader, quiet race to build the infrastructure of a post-human web. We are already seeing the emergence of machine-to-machine payment protocols, such as the Machine Payments Protocol (MPP) developed by Stripe and Tempo, designed explicitly for AI agents to transact with one another. Similarly, financial data networks like Plaid have begun rolling out agent detection capabilities to secure the broader fintech ecosystem.

What we are witnessing is the necessary bureaucratization of artificial intelligence. The initial wave of generative AI was defined by open-ended exploration and creative disruption. The next wave—agentic commerce—will be defined by rules, risk models, and liability frameworks. For AI to truly become the engine of modern commerce, it must graduate from being a digital novelty to a verified, trusted participant in the global economy. The tools to verify these invisible shoppers are finally being built, and for the merchants watching their checkout data, they cannot arrive soon enough.

Topics & Related

Event:
Partnership
Theme:
Agentic AI
Sector:
Fintech
Payments

📝 This article is still being updated

Are you a relevant expert who could contribute your opinion or insights to this article? We'd love to hear from you. We will give you full credit for your contribution.

Contribute Your Expertise →
UAID: 55316