Monica Eaton, Founder and CEO, Chargebacks911 – Interview Series

Monica Eaton is the Founder and CEO of Chargebacks911 and Fi911, as well as Chief Information Officer of Global Risk Technologies. Monica has worked tirelessly to educate merchants and financial institutions about hidden threats in the rapidly changing payment fraud landscape.

Chargebacks911 is a global fintech company specializing in chargeback prevention, dispute resolution, and first-party fraud mitigation. The company provides SaaS-based technologies and data-driven services that help merchants reduce chargeback losses, prevent fraud, and improve revenue retention. In response to Visa’s integration with ChatGPT, the company has highlighted a potentially significant challenge for the payments industry: existing dispute frameworks were built around the assumption that a consumer directly made a purchase. As agentic commerce expands, merchants, issuers, and payment networks may face growing difficulties in proving authorization, establishing intent, and determining liability when consumers dispute transactions initiated by AI agents acting on their behalf.

Fi911 provides financial institutions with back-office automation and payment operations technology designed for issuers, acquirers, banks, fintechs, and payment providers. Its platform includes tools for merchant onboarding, risk management, reconciliation, and dispute processing, with its DisputeLab™ solution focused on streamlining and automating the chargeback lifecycle. As AI-powered purchasing becomes more common, Fi911 argues that financial institutions will need more sophisticated systems for verifying consumer authorization, allocating responsibility, and resolving disputes. The company sees the rise of agentic commerce as a catalyst for modernizing dispute management infrastructure and improving transparency across the payments ecosystem.

Visa’s integration with ChatGPT is pushing agentic commerce forward. From your two decades in dispute resolution, what is the single biggest assumption in current chargeback frameworks that agentic commerce immediately invalidates?

The entire dispute framework rests on one foundation: the cardholder made the purchase. Every rule, every evidence standard, every liability allocation in the chargeback ecosystem was built around that assumption. Agentic commerce invalidates it immediately.

When a consumer delegates purchasing authority to an AI agent, the transaction is no longer a direct expression of cardholder intent. It is the output of a set of instructions, parameters, and machine decisions that may or may not reflect what the consumer actually wanted in that moment. That distinction matters enormously when a transaction is challenged. The question shifts from “did the cardholder authorize this payment?” to “did the cardholder authorize the agent to make this kind of decision?” Those are very different questions, and existing frameworks have no mechanism to answer them.

You’ve argued that the payments industry is focused on how AI completes transactions rather than what happens when those transactions are challenged. Why do you believe the dispute and liability side of agentic commerce is being overlooked?

Because the incentives in payments have always favored conversion over resolution. The companies driving agentic commerce, including networks, AI providers, and technology platforms, are measured on transaction volume, adoption rates, and consumer experience. Dispute infrastructure is unglamorous by comparison. It sits at the back end of the payment lifecycle and tends to be treated as someone else’s problem until it becomes everyone’s problem.

We have seen this pattern before. Buy now, pay later scaled rapidly, and the dispute and fraud implications were addressed years after the fact. Agentic commerce is moving faster, and the gap between front-end innovation and back-end readiness is wider than it has ever been.

In a world where consumers delegate purchasing authority to AI assistants, how should the industry redefine concepts such as intent, authorization, and accountability?

These three concepts need to be separated rather than treated as a single question because they operate at different points in the transaction lifecycle.

Intent needs to be captured at the point of delegation. What did the consumer actually instruct the agent to do, and under what conditions? That record becomes the foundation of any future dispute.

Authorization needs to be verified at the point of transaction. Did this specific purchase fall within the scope of authority the consumer granted? Spending limits and merchant category restrictions help, but they are blunt instruments. The industry will need more granular frameworks.

Accountability needs to be allocated in advance, not determined after a dispute is filed. Right now, nobody has formally agreed on who bears liability when an AI-initiated transaction is challenged. That ambiguity will create risk for every participant in the ecosystem unless accountability is clearly defined upfront.

What new forms of disputes do you expect to emerge as AI-powered purchasing becomes more mainstream through integrations like Visa and ChatGPT?

Several categories will emerge that current frameworks simply do not recognize.

Scope disputes will be common. Consumers may argue that an agent exceeded its authority even when the transaction was technically within the parameters set. “I authorized the agent to buy flights, not upgrade me to business class” is a dispute type that does not exist today.

Instruction disputes will follow. Consumers may claim the agent misinterpreted their preferences or acted on outdated instructions. These cases will be extremely difficult to adjudicate because the evidentiary trail sits inside AI systems that merchants have no access to.

Delegation disputes will be the most complex. These are situations where a consumer claims they did not fully understand the authority they were granting when they set up the agent. This is the friendly fraud vector that should concern the industry most because it is almost impossible to disprove.

If an AI assistant makes a purchase that a consumer later disputes, who should bear responsibility: the consumer, the AI provider, the merchant, the issuer, or the payment network?

The honest answer is that the industry has not resolved this yet. That unresolved question is precisely why the dispute infrastructure needs to be addressed before agentic commerce reaches mass scale.

My view is that responsibility should follow the audit trail. If a consumer set clear parameters and the AI acted within them, the consumer bears primary accountability. If the AI acted outside its authorized scope, responsibility shifts toward the provider. If the merchant had no reasonable way to verify authorization, they should not be the party absorbing the loss, which is what the current chargeback model would effectively produce.

What I am confident about is that the existing default, where merchants bear the cost of disputed transactions, was not designed for this environment and will produce deeply unfair outcomes if applied without modification.

Through Chargebacks911, you’ve helped businesses combat friendly fraud for more than a decade. How do you expect first-party fraud to evolve when consumers can claim that an AI agent, not them, made the purchasing decision?

It creates a significant new avenue for abuse, and I want to be direct about that rather than soften it.

Friendly fraud has always relied on plausible deniability. The consumer claims they did not make a purchase, did not receive goods, or did not recognize a charge. Agentic commerce adds a new layer of deniability that is far harder to challenge. “The AI did it, not me” is a defense that sounds reasonable to a consumer, to an issuer’s dispute team, and potentially to a regulator.

Historically, fraud follows adoption. As agentic commerce scales, bad actors will look for ways to exploit the ambiguity surrounding AI-initiated purchases. That gives the industry a limited window to build the right infrastructure before losses begin to rise.

What new data, audit trails, or verification mechanisms will merchants and financial institutions need to prove valid authorization in AI-driven transactions?

This is where the operational work needs to happen, and it is significant.

Merchants will need access to delegation records. They will need evidence showing what authority the consumer granted the agent, when they granted it, and under what conditions. Today, that information sits with the AI provider, and there is no standardized mechanism to share it in a dispute context.

Issuers will need transaction-level records that go beyond the payment itself. They will need visibility into what instructions triggered the purchase, what the agent’s decision logic was, and whether the transaction fell within the consumer’s stated parameters.

This is precisely why the industry needs modern evidence-management and dispute-resolution platforms. Systems like UDMS and ResolveLab were built around the idea that dispute decisions should be driven by richer data, stronger audit trails, and greater transparency. But those platforms can only work with data that exists and is accessible. Right now, the raw evidence infrastructure for agentic commerce has not been defined.

Today’s chargeback frameworks were designed around human purchasing behavior. What are the biggest gaps in existing dispute-resolution systems when applied to agentic commerce?

Three gaps stand out.

Evidence standards were designed for human transactions. Existing reason codes, representment processes, and documentation requirements reflect a world where a person chose to buy something. None of them account for machine-initiated purchases, delegated authority, or AI decision logs.

Liability allocation is binary in a way that does not fit agentic commerce. The current model assigns liability to either the merchant or the issuer based on relatively simple criteria. Agentic commerce introduces multiple parties, including the consumer, AI provider, network, merchant, and issuer. The existing framework has no mechanism to apportion responsibility across them.

Timelines are misaligned. Chargeback windows were designed around human behavior and human memory. In an environment where AI agents may be making dozens of purchases on a consumer’s behalf, the gap between transaction and dispute will widen, and the existing timeframes will need revisiting.

Do you foresee new industry standards or regulations emerging around AI-generated purchases, and what role should payment networks, regulators, and fintechs play in establishing them?

Standards will emerge. The question is whether they emerge proactively or reactively. Historically, standards often emerge after new forms of risk become visible at scale. The industry should be working now to establish voluntary frameworks before regulators are forced to act.

Payment networks are best positioned to move first. They sit at the center of the transaction lifecycle, they have relationships with all parties, and they have the technical infrastructure to enforce standards at scale. Visa’s announcement demonstrates that the industry is thinking seriously about the front end of agentic commerce. The same rigor now needs to be applied to dispute and liability infrastructure.

Regulators in both the United States and the United Kingdom are already paying close attention to AI in financial services. The frameworks that emerge from bodies such as the CFPB and the FCA will likely focus on consumer protection first. That means merchants and issuers need to be actively engaged in shaping those standards rather than responding to them after the fact.

Looking five years ahead, what does a well-functioning agentic commerce ecosystem look like, and what steps should the payments industry take today to ensure innovation doesn’t outpace accountability?

A well-functioning agentic commerce ecosystem is one where the speed and convenience of AI-driven purchasing is matched by an equally sophisticated infrastructure for establishing intent, resolving disputes, and allocating liability when things go wrong. Those two things need to develop in parallel. Right now, they are not.

The steps the industry should take today are specific. First, establish standardized delegation records: a consistent format for capturing what authority a consumer has granted an AI agent that is accessible to all parties in a dispute. Second, define liability allocation frameworks before disputes start appearing at scale, not after. Third, invest in the evidence infrastructure, including the audit trails, verification mechanisms, and dispute-management systems that agentic commerce will require.

The companies and networks that do this work now will define how agentic commerce operates at scale. The ones that do not will spend the next decade trying to retrofit accountability into systems that were never designed for it.

Thank you for the great interview, readers who wish to learn more should visit Chargebacks911 and Fi911.