AI shopping agents are being wrongly flagged as fraud, costing merchants legitimate revenue
Chargebacks911, a global leader in dispute resolution and chargeback prevention, is warning that the rapid growth of agentic commerce is creating a significant and largely overlooked risk for merchants: legitimate AI-initiated purchases being misclassified as fraudulent bot activity, triggering false declines and revenue loss at scale.
The warning comes as agentic commerce, in which AI systems autonomously search, compare, and complete purchases on behalf of consumers, moves rapidly from concept to commercial reality. According to The Paypers Global Ecommerce Report 2026, agentic commerce could account for 25% to 30% of global online purchases by 2030. Visa and Mastercard are already piloting agent-initiated transactions with major banking partners, and platforms including Perplexity, Walmart, and Amazon are enabling AI agents to transact directly on consumers’ behalf.
Yet as this shift accelerates, fraud detection systems have not kept pace. According to Imperva’s 2025 Bad Bot Report, 51% of internet traffic is now generated by bots, of which 37% is considered malicious. Historically, merchants could distinguish between automated traffic and genuine consumer behaviour. That distinction is rapidly disappearing. Today’s AI shopping agents operate within browsers, generating traffic patterns that appear increasingly human, and in doing so are triggering fraud systems designed for a world in which a human being was always behind the transaction.
Chargebacks911 says the implications for merchants are twofold. While industry attention has focused primarily on the risk of disputed AI-initiated transactions, the reverse problem, blocking legitimate agentic purchases before they complete, represents an equally significant and more immediate threat to revenue. A false decline carries no chargeback, but the cost in lost sales, damaged brand trust, and reduced visibility to AI agents is immediate and growing.
“The fraud systems most merchants rely on today were built to detect bad human behaviour. They were not designed for a world where a legitimate AI agent and a malicious bot look almost identical,” said Monica Eaton, founder and CEO of Chargebacks911. “As agentic commerce scales, merchants face a clear choice: adapt their detection and evidence infrastructure now, or watch a growing share of legitimate revenue get declined by their own systems.”
Traditional fraud prevention relies on behavioural signals tied to human interaction, including device fingerprints, session patterns, click sequences, and authentication flows. Agent-initiated transactions disrupt each of these markers. Without a clear evidence trail showing what the agent was authorised to do, what it actually executed, and on whose behalf, merchants lack the data needed to accurately classify the transaction or to defend against a dispute if the purchase is later challenged.
Chargebacks911 addresses this through its Unified Dispute Management System (UDMS) and ResolveLab, which use AI and machine learning to build and analyse the evidence architecture that agentic transactions require. Rather than relying on point-of-transaction signals alone, UDMS captures the full consent and permission trail, including what the agent was authorised to do, the limits in place, and a timestamped record of each action taken. This gives merchants and financial institutions the visibility needed to distinguish a legitimate agent transaction from malicious automated activity, and to act with confidence in either direction.
“In an agentic commerce environment, the evidential anchor shifts from a real-time human action to a prior consent framework,” said Donald Kossmann, chief technology officer at Chargebacks911. “Merchants need systems that can read that framework accurately and quickly. The organisations that build that capability now will not only reduce false declines; they will have a structural advantage as AI-driven purchasing becomes the norm.”
Chargebacks911 recommends three immediate actions for merchants: establish granular permission frameworks for AI agents that transact on their platforms; invest in evidence capture infrastructure that logs agent authorisation alongside transaction data; and review fraud detection thresholds and rules to account for the behavioural differences between human and agent-initiated purchases. Merchants operating internationally should ensure these frameworks reflect the varying pace of agentic commerce adoption across markets including the US, UK, and key APAC and LATAM regions, where consumer appetite for agent-assisted purchasing is accelerating rapidly.
“The industry has rightly focused on what happens when an AI agent makes a purchase the customer did not want. The question that remains largely unasked is what happens when a merchant’s fraud system refuses the purchase the customer did want,” said Eaton. “Both problems need solving, and both require the same thing: a clear, auditable record of what was authorised and what happened.”