- 73% of software providers and 65% of merchants are actively building or planning agentic commerce capabilities (2026 Payments Report).
- 43% of fraudulent chargebacks surged over the past year, up from 29%.
- 81% of ISVs plan to use payment processing revenues to subsidize AI compute costs.
Experts agree that agentic commerce is rapidly advancing without adequate governance, leaving merchants vulnerable to financial risks and forcing them to absorb hidden AI-related costs.
Agentic Commerce's Hidden Tax: Who Pays When AI Goes Rogue?
MADISON, WI – November 17, 2026 – We have officially crossed the threshold from artificial intelligence that talks to artificial intelligence that buys. Agentic commerce—where autonomous algorithms initiate transactions, negotiate prices, and execute payments on a consumer's behalf—is rapidly rewriting the architecture of modern retail. Yet, as the technology accelerates, the financial and regulatory guardrails meant to protect the ecosystem are dangerously lagging behind.
According to the newly released 2026 Payments Report by Wind River Payments, a Wisconsin-based merchant services provider, 73% of software providers and 65% of merchants are actively building or planning to introduce agentic commerce capabilities. However, this gold rush is built on a fragile foundation. Only 52% of software providers believe current payment rules, chargeback processes, and fraud liability frameworks are clear enough to govern these machine-initiated transactions.
This disconnect reveals a profound tension at the intersection of modern culture and commercial strategy: the consumer desire for frictionless, delegated purchasing is colliding with a massive governance void. And behind the scenes, the staggering cost of building these autonomous agents is quietly reshaping the economics of software, with merchants footing the bill.
The Governance Void: When the Machine Makes a Mistake
When a human consumer clicks "buy," the chain of authorization is linear and legally established. When an AI agent, given a broad mandate to "find the best deal on a winter coat," purchases the wrong item, exceeds a budget, or falls victim to a sophisticated scam, the chain of liability shatters.
This "liability gap" is currently the most significant blindspot in fintech. The Wind River Payments research indicates that 43% of independent software vendors (ISVs) cite unclear liability for AI-driven fraud or errors as a primary commercial challenge.
Major card networks are scrambling to patch the holes, but their approaches are deeply fragmented. In April of this year, Visa updated its Core Rules to explicitly route AI-initiated purchases into existing card-not-present dispute processes, essentially decreeing that the cardholder is responsible for any actions taken by an agentic payment provider. Mastercard has launched "Agent Pay" and is developing a "Verifiable Intent" specification to create tamper-resistant records of consumer instructions, but internal network documentation still concedes that dispute liability remains ambiguous.
American Express has taken a slightly different route, launching "Agent Purchase Protection" to cover eligible cardholders from charges resulting from AI errors, provided the purchase intent is authenticated. However, this framework protects the consumer while leaving the merchant exposed to the ultimate financial loss of the chargeback.
"Software providers are feeling pressure from multiple directions," said Tyler Kattre, President at Wind River Payments. "Features are easier to replicate, the pressure to innovate is growing, and agents are changing how purchases take place. Merchants face more sophisticated fraud, rising costs, and shifts in how customers find and pay their business. Payments is where these pressures meet, and the strongest software provider-merchant partnerships will be better positioned to navigate this environment and come out ahead."
The Merchant's Dilemma: Fraud Outpacing Defense
The ambiguity of card network rules is compounded by a profound regulatory lag. Foundational consumer protection laws in the United States, such as Regulation E (Electronic Fund Transfer Act) and Regulation Z (Truth in Lending Act), were constructed for a binary world: either a human authorized a transaction, or they did not. These regulations struggle to define "unauthorized" transfers when dealing with delegated machine authority.
As one payments legal expert noted, financial institutions are delaying widespread adoption of agent-initiated payments until clearer frameworks exist, fearing the legal exposure of the current wild west.
In the absence of clear federal guidance, merchants are absorbing the impact. The report highlights that fraudulent chargebacks surged to 43% over the past year, up from 29%. Furthermore, 62% of software providers warn that fraud sophistication is actively outpacing their defenses.
When asked who should be financially liable when an AI platform runs a fraudulent purchase, merchants are sharply divided. While 34% believe the AI platform operating the agent should pay, 16% point to the software provider, 16% blame the consumer, and 8% accept that the merchant must bear the cost. This lack of consensus is a ticking time bomb for B2B relationships across the retail sector.
The Hidden AI Tax: Subsidizing the Compute Bill
Beyond the unresolved questions of liability, agentic commerce is triggering a seismic shift in how business software is monetized. The development of large language models and autonomous agents requires staggering investments in computing power and specialized talent.
The survey reveals that 91% of software providers saw their AI-related costs increase this past year, forcing 30% of them to scale back their ambitions due to infrastructure constraints. To survive, these platforms are turning to a stealthy monetization strategy: the hidden AI tax.
Rather than imposing massive subscription price hikes—which directly alienate users and drive churn—software vendors are embedding the cost of AI into the transaction itself. A striking 81% of ISVs report they plan to use payment processing revenues, markups, or payment-related fees over the next 12 months to subsidize their escalating compute expenses.
This represents a fundamental evolution in embedded finance. Historically, SaaS platforms integrated payments to reduce friction and create a sticky user experience. Today, payment processing is being weaponized as a primary funding mechanism for artificial intelligence research and development. Nearly a third of merchants using integrated payments report that their software provider has already increased fees, explicitly attributing the hike to higher AI development costs.
For major integrated platforms, this strategy allows them to fund technological advancements without the friction of direct subscription increases. But for the small to mid-sized merchant, it acts as an unavoidable toll on their daily revenue stream. More than a third of these merchants state they will be forced to pass these increased processing costs directly onto the end consumer through surcharges.
The Race for Verifiable Intent
As AI-driven traffic to retail sites continues to experience exponential, multi-thousand-percent year-over-year growth, the industry is frantically trying to build the airplane while it is already in the sky.
The focus across the payments ecosystem has acutely narrowed onto the concept of "intent." The goal is to mathematically prove that an AI agent's action perfectly aligns with the cardholder's original, specific instruction. If intent can be verified and cryptographically sealed before the transaction is processed, networks believe they can mitigate the risk of rogue purchases and streamline the inevitable wave of agent-driven disputes.
Yet, technology alone cannot solve a legal and commercial standoff. Until regulators modernize consumer protection laws to account for autonomous machine purchasing, and until card networks establish a unified, transparent liability framework that doesn't simply default to punishing the merchant, agentic commerce will remain a high-stakes gamble. The consumer may enjoy the luxury of having an AI personal shopper, but the hidden taxes and unresolved risks of that convenience are currently being paid by the businesses sitting on the other side of the screen.
Topics & Related
Agentic AI
Payments
AI & Machine Learning
📝 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 →