Why agentic commerce is creating new risks for merchants

The rapid emergence of agentic commerce—where AI tools assist or act autonomously in consumers’ online purchases—is reshaping ecommerce faster than most merchants can adapt.

According to new research from fraud prevention firm Riskified, nearly three-quarters of global shoppers (73%) are already using artificial intelligence at some stage of their shopping journey.

Yet behind the convenience lies a growing threat: AI-driven purchases blur accountability, complicate fraud prevention, and expose merchants to fresh forms of financial and reputational risk.

AI is Redefining the Consumer Journey

Agentic commerce creating new risks for merchants

AI assistants such as ChatGPT, Gemini, and Copilot are quietly becoming embedded in the retail experience.

Riskified’s survey of more than 5,000 consumers reveals that 45% use AI to generate product ideas, 37% to summarise reviews, and 32% to compare prices.

Although only 13% have completed a purchase based on an AI recommendation, 70% say they would be comfortable allowing an AI agent to make purchases on their behalf.

This comfort level signals the dawn of the first truly AI-powered holiday season. Over half (58%) of respondents expect to use AI assistants for gift shopping this year.

However, this shift also means that a growing number of transactions will originate outside traditional merchant channels—without the consumer ever directly visiting a retailer’s website.

A New Type of Risk Exposure

The convenience of agentic commerce introduces uncertainty over who bears liability when disputes arise.

As Jeff Otto, Chief Marketing Officer at Riskified, explains, “When an AI shopping agent makes a purchase, is the merchant still liable when a dispute occurs—even though the shopper never visited the merchant’s site during checkout?”

That uncertainty is fertile ground for fraudsters.

Organised fraud rings, already equipped with advanced AI tools, are likely to exploit weaknesses in identity verification, account recovery, and chargeback systems.

A transaction made via an AI assistant could be disputed by a legitimate customer claiming their bot made a mistake—or could stem from a hijacked AI account.

Either way, merchants are left shouldering the cost and complexity of chargebacks.

Consumer Trust vs. Merchant Control

Consumers are divided between excitement and unease.

Payment security ranks as the top concern for 32% of respondents, followed by privacy (26%), potential errors (18%), and loss of control (17%).

Yet trust in AI is rising fast: 36% of consumers say they trust AI to influence their purchases—almost matching the 38% who rely on in-store sales associates. Only a quarter prefer to shop online without AI assistance.

Building Guardrails for the Agentic Era

As the rollout of the Agentic Commerce Protocol (ACP) and others accelerates—an open standard allowing AI systems to complete purchases directly—fraud prevention teams are being urged to take a leadership role.

Riskified recommends three strategic imperatives:

• Educate leadership on the risk-reward trade-offs of agentic commerce.

• Advocate for data transparency from AI platforms to ensure merchants can make informed fraud decisions.

• Leverage shared intelligence by collaborating with networked fraud detection ecosystems.

Otto argues that fraud teams are uniquely positioned to guide this transition: “Their expertise isn’t about saying ‘no’ to innovation—it’s about building the guardrails that allow companies to say ‘yes’ confidently.”

As merchants prepare for a future in which AI acts as both customer and risk vector, the challenge will be balancing growth and innovation with accountability and trust.

In the agentic commerce era, success will hinge on collaboration—between merchants, regulators, and AI platforms—to ensure that the next wave of digital retail remains both intelligent and secure.