IXOPAY webinar takeaway: preparing for the trust challenges of autonomous commerce

Paula Albu, Junior Editor at The Paypers, shares the key takeaways from the webinar co-hosted with IXOPAY and Nuvei, titled ’Preparing for the Trust Challenges of Autonomous Commerce’.

In this webinar, Jill Willard, CTO at IXOPAY, and Mentxu Triviño, VP Payment Partnerships at Nuvei, joined moderator Tracy Brown, Fraud and Payment-risk Expert, to examine what the rise of agentic payments means in practice - from the integrity of the existing four-party model to how liability, fraud signals, and customer controls will need to evolve. 

Autonomous commerce is no longer theoretical. AI agents can shop, decide, and pay on a customer’s behalf. Their transactions are largely invisible, flowing through a payment infrastructure that was never designed to accommodate them. While technology is built to be adaptive, the real question all parties face is how to build trust when the buyer is not human.

From fraud detection challenges to liability frameworks and customer control mechanisms, the conversation revealed that while the technology and demand are ready, the execution, governance, and trust infrastructure are not – at least, not yet.

A telling poll of webinar attendees illustrated exactly where the industry stands. 70% of participants had never used an autonomous agent, and only 27% had. The humour appeared in the remaining responses, with 3% suspecting an agent had already stolen their card details, a response that captured the ambient anxiety surrounding a technology most people have heard about, but few have actually engaged with on a payment level.

Here are the key takeaways of the webinar. 

A new layer on top of existing complexity

The traditional four-party payment model (issuer, acquirer, merchant, and card network) is now having a fifth participant: the AI agent. As Mentxu framed it, this is not simply about adding complexity; it is about building the right governance framework with clear rules and responsibilities for each participant. Merchants must retain control over their products, pricing, and business rules, while payment providers serve as the trust layer, ensuring data flows transparently across the entire transaction chain.

One thing is certain: payment has always been complicated, and there is a constant need to manage complexity, including fragmented PSPs, multiple acquirers, and a range of alternative payment methods, each with its own rules and risk signals. As Jill noted, payment orchestration, the layer that creates a single decision framework across all those inputs, was already solving for that complexity. Agentic transactions add another layer, but the fundamental challenge remains the same. The key difference is that agent transactions now represent a new channel with distinct risk profiles, and merchants need the visibility to treat them accordingly, rather than having them co-mingled with standard e-commerce traffic.

For payments and fraud leaders already managing routing logic and approval rate optimisation across multiple flows, the setup is largely familiar. What is changing are the inputs. 

All things fraud

One of the more operationally pointed observations in the session concerned what is already happening to fraud and risk models. Agent-initiated transactions are currently flowing through as standard e-commerce, because the infrastructure does not yet have the visibility to distinguish them. The problem is that bot behaviour differs materially from human behaviour: it is more linear, faster, and less exploratory, and traditional risk systems flag it as atypical. The result? Elevated decline rates on what may be entirely legitimate, consumer-authorised transactions.

Jill emphasises that having more context and data is essential, as fraudsters adapt quickly to the environment. Mentxu added that while the ecosystem has developed strong capabilities for detecting fraud based on human behaviour, agent-based fraud represents a new frontier, and the identity verification framework needs to evolve rapidly. Two distinct challenges emerged from the discussion: determining whether an agent is good or bad, and, separately, assessing whether its behaviour is consistent with expectations. The industry needs to address both.

For fraud and payments teams, this creates an immediate operational issue. Agentic traffic, predicted to grow, will begin distorting monitoring program metrics computed against entirely different behavioural baselines. Anomaly detection around chargebacks and other indicators needs to be reviewed as soon as possible, so that teams understand what their current data actually reflects.

AI - faster than standardisation?

Tracy raised an important question on everyone’s lips: if AI advances faster than other implementations, what progress will it make? As Jill noted, the greatest thing about standards is that there are always plenty of them. The industry is constantly in a discovery phase, with various frameworks competing for dominance, none of which has yet achieved widespread adoption.

Drawing a parallel to the evolution of e-commerce, which took several years to develop controls such as 3D Secure, tokenization, CAPTCHA, and network tokens, both speakers cautioned against expecting overnight standardisation. Triviño advised merchants not to chase every new protocol, but instead to partner with payment providers who can navigate the evolving landscape without requiring merchants to rebuild their infrastructure each time a new standard emerges. As a merchant, understanding where you stand and choosing partners who align with evolving standards are not optional; they are essential.

Who owns the risk?

Perhaps the most substantive thread of the discussion centred on liability. When an AI agent executes a transaction autonomously, who owns the dispute? The consumer who delegated authority? The agent provider? The merchant who accepted the transaction?

For now, the answer is clear and uncomfortable for merchants. Liability falls primarily on the merchant across most existing protocols. The more complex scenario arises when the customer is not present, when an agent acts overnight, for instance, and at that point, the liability question becomes genuinely unsettled. 

A notable development on this front is American Express recent launch of the ACE Developer Kit and Amex Agent Purchase Protection, a commitment to safeguarding eligible customers and merchants from charges resulting from AI agent errors. As Jill noted, Amex’s closed-loop model offers unique visibility that other card networks do not yet have.

Mentxu offered a useful distinction between the current phase and what follows. Today, agents are largely performing the discovery phase, researching, comparing, and surfacing options, while the actual payment remains a human-confirmed action in the merchant environment. True agentic payment execution, where the agent completes the transaction without the consumer present, is the next phase and the one in which liability frameworks will need to be renegotiated across payment networks.

Customer control and the agent wallet

On the customer side, the session examined how people will be expected to manage agents acting on their behalf, and whether the necessary infrastructure exists to support that.

A live poll during the webinar revealed that 50% of attendees would trust an AI agent to make payments only if they could review and approve transactions, while 17% would clear spending limits and controls, and 14% required stronger fraud-protection guarantees. A notable 19% said they would not trust an agent at all today.

Willard described the emerging concept of the agent wallet: a dedicated credential store provisioned specifically for agent use, with spending limits, merchant category restrictions, and card-level controls defined by the customer. The analogy is to network tokens provisioned on a mobile device, but applied to an agent rather than a phone. Moreover, the agent approval model is likely to follow a pattern similar to 3D Secure authentication flows, with optimal processing for low-risk transactions and a triggered review for higher-value or higher-risk decisions.

Mentxu carefully separated customer preference from consumer behaviour, noting that adoption will follow trust, and trust will follow positive experiences. The analogy she drew was to digital wallets: customers trusted Apple Pay and Google Pay not because they understood the cryptographic mechanisms, but because they trusted the brands. Agent commerce will likely follow the same pattern, growing as the customer experience becomes reliable and the guardrails become legible.

What comes next, and what to do now

Both speakers converged on a similar set of near-term recommendations for payment professionals, framed around a crawl-walk-run logic. Willard advised teams to begin by understanding their payment data in detail: monitoring chargebacks, establishing auditability and traceability across the transaction layer, and defining the level of risk they are willing to accept before choosing protocols.

Triviño’s framing was complementary: merchants should be able to delegate tasks to an agent without delegating the risk and control. The payment infrastructure underneath must be scalable, reversible, and compliant, but not just with network policies, but also with local regulations, because payments remain fundamentally local even as the technology becomes global.

The standards are being shaped now, before the frameworks are set. The payment professionals best positioned as agentic commerce matures will be those who are already in those conversations, rather than those receiving the outcomes.

The message is clear: don’t wait for the standards to be finalised, start preparing today because the agents are already at checkout.

This webinar recap only highlights the key points of the discussion. For the complete take on what merchants must do to prepare for an increasingly agent-driven future, including the insightful Q&A session at the end, watch the webinar recording here.

About the author

 

Paula Albu has experience in content writing and editing, as well as being a creative storyteller. As a Junior Editor at The Paypers, she investigates Web3 technologies along with the latest trends and regulations in banking and fintech. Paula is committed to turning complex industry topics into engaging, accessible content that resonates with readers and creates a meaningful connection. She is available via LinkedIn or at paula@thepaypers.com.

 

About IXOPAY

IXOPAY is the enterprise-grade global payment infrastructure platform built for the era of agentic commerce, equipping merchants and businesses with AI-driven intelligence, orchestration, advanced tokenization, and the tools to power every step of their payments journey. Learn more at https://www.ixopay.com/.