The future where AI agents get more room to pay for products and services seems to be getting nearer. Most fintech providers are already preparing common protocols, controls, and payment rails that will help agentic commerce bloom full-scale. Don’t we forget anything? Whether agentic or not, commerce remains basically sales, which, besides purchases, also involve returns and payment disputes. Are fintechs ready to handle that aspect when they have to deal with an AI bot rather than a human customer?
What are payment disputes (chargebacks)?
In commerce, sellers often face chargeback disputes when a customer asks their bank to reverse a transaction. Here’s why it may happen:
- transaction is unauthorized (fraudulent);
- purchased item was not received for some reason;
- advertised product did not appear as described;
- customer doesn’t recognize the charge (although it might be legitimate);
- the payment is unwanted but scheduled, e.g. occurred due to subscription cancellation issues or “forgotten” recurring payments;
- customers are falsely disputing charges or exploiting refund policies in what’s called first-party fraud incidents (most common fraud type, BTW).
When the bank receives such a request, it may temporarily refund the customer while investigating or just freeze the funds in an escrow account. The merchant, who’s informed accordingly, must either accept the loss or provide evidence to defend the transaction.
Would the rise of agentic commerce necessarily trigger dispute rise?
The vision of agentic commerce is beautiful. However, at the onset (where we are now), it looks messy. Just remember the scandal with Amazon AI shopping tool testing occurring a few months ago. The experiment was a small-scale, but dozens of small retailers discovered their products were listed and sold on Amazon without their consent. Some products were not even on sale (out of stock). As many mismatches appeared, customer complaints and refund requests were soon to follow.
Should this inevitably happen to any e-commerce firm deciding to embark on an agentic AI purchase journey? Not necessarily, but quite possibly, especially at an early tech deployment stage. The main potential friction points are:
- ambiguous consent practices and lack of user control;
- expectation gaps between what AI agent “thinks best” and what actual customer subjectively prefers;
- subscriptions and recurring payments, which are already biggest chargeback source, shall not change their nature after agentic automation;
- customers who practice first-party fraud already may be tempted to disguise that as AI agent “making mistake”.
Who’s going to be accountable for dispute increase and how to manage that?
When a transaction is triggered via an agent platform, it becomes less clear who is responsible for any dispute arising: the merchant, the agent provider, or the payment processor. Those boundaries need to be set clearly before we proceed to full-scale agentic commerce. The absence of transparency in such discussion points is one of the factors making people cautious about AI-driven shopping.
While the question hangs in the air, the agentic commerce boom is just around the corner. One clear illustration is Perplexity’s use case. The AI firm’s estimated annual recurring revenue rose to more than $450 million in March, 50% growth in a month, just as it shifted from search to AI agents.
Monica Eaton, Founder & CEO, Chargebacks911, commented:
“Perplexity’s revenue surge is a clear signal that agentic commerce is moving from concept to commercial reality. When a business reaches an estimated $450 million in annual recurring revenue, driven in part by AI agents carrying out tasks on users’ behalf, it reflects a structural shift and not merely a passing trend. However, every agent-initiated transaction that scales is also a potential dispute that scales. Merchants typically cannot generate or access the level of insight required to manage this exposure independently, and many rely on third-party providers to gain the visibility needed to respond effectively.
The commercial case for agentic commerce is now clear, but the supporting dispute infrastructure still has ground to cover. As AI agents increasingly act on behalf of consumers, real-time insight into dispute data becomes a baseline operational requirement. Businesses must continuously measure performance and respond to live data if they are to remain competitive in this environment.”
Once the infrastructure for agentic disputes is ready-made and provides real-time data insights, the situation with chargebacks may actually improve.
For once, well-trained agents can reduce human errors (e.g. wrong item, wrong address), which are a common dispute trigger. Agents can also confirm preferences, budgets, and purchase constraints before transaction occurs — something humans often skip. Finally, AI-mediated transactions would include detailed and transparent logs, explaining why, where and how this purchase was made, potentially strengthening evidence in dispute resolution and helping customers better recognize the transaction.
Brief summary
Agentic commerce is rapidly moving toward the mainstream, but dispute management remains an open challenge. Payment disputes (chargebacks) arise from fraud, unmet expectations, or subscription issues: risks that may initially grow as AI agents make purchasing decisions. Ambiguous consent, expectation gaps, and unclear liability across merchants, platforms, and processors add complexity. While early adoption may increase disputes, improved controls, transparency, and AI-driven insights could ultimately reduce them by minimizing human error and strengthening transaction accountability.
