Agentic AI Needs Strong Verification to Unlock Safe, Scalable Commerce

Watch more: Agentic AI Needs Strong Verification to Unlock Safe, Scalable Commerce

For merchants, the coming wave of agentic artificial intelligence (AI) represents both a revenue opportunity and a rising fraud risk.

Artificial intelligence (AI)-powered shopping assistants—algorithms executing purchases on behalf of consumers—are already starting to interact with checkout systems. But without clear verification, retailers face uncertainty over whether they are transacting with a legitimate agent operating with consumer consent or with a malicious bot attempting fraud.

“Agents are already shopping on merchant websites in ways that aren’t being identified,” said Cindy Turner, chief product officer at Worldpay. Transactions may look identical to a normal purchase, but when fraud or disputes arise, merchants are left holding liability.

The challenge is heightened by the fact that merchants have spent years investing in fraud tools designed to block bots entirely. Now, the introduction of AI-powered agents means some of those bots may represent real consumer intent—and lost sales if blocked.

Balancing fraud control with revenue growth is becoming a critical need for merchants navigating this evolving landscape.

Despite the attention, agentic commerce is still in its early stages. Execution remains limited to small pilots rather than industry-wide adoption. 

“We’re at a huge pivot point where there’s lots of experimentation but nothing yet at scale,” Turner said. Different models are being tested, from headless agents making purchases directly on merchant sites to large-language models experimenting with merchant-of-record approaches. But right now, there is no standard way for merchants or issuers to validate agent transactions and manage liability.

This lack of standardization creates uncertainty for all parties in the payments chain. For issuers, agent-based purchases appear no different from standard card transactions, complicating dispute resolution. For merchants, the absence of transparent consent verification increases chargeback exposure when consumers argue over what was authorized.

To provide clarity, Worldpay and Trulioo announced earlier this month that they are teaming up to introduce identity safeguards for agentic AI transactions. The cornerstone is the Know Your Agent (KYA) framework, underpinned by a tamper-proof Digital Agent Passport that establishes a verifiable trust chain from consumer to agent to merchant.

For Worldpay, which processes more than $2.5 trillion in merchant volume annually, the partnership provides a standardized framework for its massive global merchant base. There’s a normalization of API data so merchants consistently receive consent information from agents, Turner said, while Worldpay already provides the tokenization and credential storage to keep payments secure within that flow. 

Trulioo, a global leader in identity verification, brings the digital identity and compliance layer. Its platform can verify consumer and business identities across 195 countries, supporting Trulioo’s role as the arbiter of agent trustworthiness. Vicky Bindra, CEO of Trulioo, told PYMNTS in the same interview that its contribution ensures merchants can distinguish legitimate agents developed by credible parties from malicious bots.

“Instead of click to buy, we are moving to ‘code to buy,’” Bindra said. “Agents are nothing more than lines of code. Making sure that code is tied to the right consumer, with clear consent parameters, is essential to building trust.”

One of the partnership’s core goals is balancing verification with seamless commerce. For merchants worried about increases in abandoned carts, both executives were quick to note that the KYA framework is designed to add trust signals without adding steps for consumers or merchants.

By embedding trust directly into the rails, “you reduce rather than increase transaction friction,” Turner and Bindra noted. Merchants can accept agent-driven purchases with greater confidence instead of fearing fraud.

Bindra emphasized that reducing risk encourages adoption: “People can move from cautious interest to confident deployment,” he told PYMNTS.

Liability transfer remains a major issue as commerce shifts to agentic models. If consumers dispute purchases initiated by their agents, merchants currently bear disproportionate exposure.

With Worldpay and Trulioo’s framework, proof of consumer consent can be cryptographically linked to each transaction, offering issuers and merchants shared transparency.

Bindra also pointed to future industry needs, such as agent certification systems, agent directories and clearer liability handoffs between merchants and issuers. “Ultimately, this will look like an app store … where people can very confidently go to the right agents,” he said.

While AI shopping assistants for retail purchases capture headlines, the near-term opportunity for agentic AI may also expand beyond that setting. Both executives highlighted early use cases where transaction parameters are simpler and well-defined. 

According to Trulioo’s consumer research, initial adoption will likely surge in bill pay, subscriptions and reorders of everyday items such as pet food. These transactions are low risk for both merchants and consumers, require little customization and are ripe for automation.

On the merchant side, Worldpay is working closely with travel providers, where Turner sees strong applicability despite higher complexity. For travel, there’s the opportunity to insert a consumer validation step before final booking, she stated. Issuers and merchants can use the same infrastructure but layer in human confirmation for high-risk, high-value categories like flights. 

In an era where “code to buy” may become as common as “click to buy,” Worldpay and Trulioo are betting that verification, not friction, will define the future of agentic AI.

As Bindra told PYMNTS, “the goal is that without adding friction or abandoned carts, you actually increase the velocity of transactions because we’ve created a trusted infrastructure for consumers and for merchants and for issuers to complete that transaction.”

The post Agentic AI Needs Strong Verification to Unlock Safe, Scalable Commerce appeared first on PYMNTS.com.