Agentic commerce boom could revive old fraud problems

Agentic commerce is quickly becoming ecommerce’s next big bet, with retailers eyeing AI agents that can search, compare, recommend, and complete purchases on a shopper’s behalf. The pitch is obvious: less friction, faster transactions, and more automation at a time when customer acquisition costs remain high and growth is getting harder to find.

That enthusiasm is understandable, but the warning from Ravelin’s CEO is straightforward: ecommerce has seen this pattern before. Online payments made shopping easier but expanded card-not-present fraud. Mobile commerce reshaped buying habits while adding new security challenges. And AI has already given fraudsters more powerful tools to scale attacks. There is little reason to assume agentic commerce will avoid the same cycle.

Retailers are moving fast because they do not want to miss another platform shift. Some see agentic commerce as potentially as significant as the move to mobile commerce itself. But hype around new technology does not settle the harder questions around risk, especially when merchants may be placing more trust in AI agents than in their own customers.

Recommended reading

Europe’s chip push won’t end US cloud reliance

According to the piece, many ecommerce businesses are already grappling with:

• refund abuse
• chargeback fraud
• promotion abuse
• policy exploitation

AI agents may look safer because they follow instructions and do not intentionally game systems for personal gain. But that does not remove risk; it changes its shape. New vulnerabilities are emerging, and many businesses are only beginning to understand them.

Liability and governance questions

One of the biggest unresolved issues is accountability. If an AI agent makes a bad purchasing decision, triggers a fraudulent transaction, or is manipulated into buying something it should not, responsibility is far from clear. The consumer, retailer, technology provider, or operator of the agent could all be pulled into the dispute.

Existing ecommerce frameworks were built around human decision-making. Agentic commerce adds a layer of autonomy that blurs those lines. For merchants, that means governance, oversight, and visibility matter before these systems are deployed at scale. Businesses need to know who is operating the agent, how it is making decisions, and what safeguards exist when something breaks.

Fraudsters are likely to adopt it fast

If previous waves of ecommerce innovation are any guide, criminals will move quickly. The article points to several likely scenarios: hijacking legitimate agents, creating fake agents that mimic trusted brands or services, and deploying autonomous agents to probe for weaknesses and exploit them continuously.

Promotions, loyalty programs, and refund systems are especially likely to be targeted. The shift could be less about a simple rise in fraud volume and more about faster, more adaptive attacks operating at machine speed. In that environment, merchants will need stronger visibility into customer and transaction behavior, more diverse data, and faster ways to detect unusual activity as threats change.