AI-Assisted Commerce is Here. Trust Will Define Who Wins

Google’s introduction of the Universal Commerce Protocol (UCP) marks a structural shift in how digital commerce operates. Launched in January 2026 and already enabling in-app purchases from Etsy and Wayfair through Google’s AI Mode in Search and Gemini (with integrations from retailers and platforms like Shopify, Target, and Walmart on the horizon), UCP is designed as a shared infrastructure layer for agent-driven shopping.

At its core, UCP standardizes how product data, pricing, carts, payments, and order tracking are structured, enabling retailers and platforms to communicate in a common language. For decades, commerce systems have been fragmented and incompatible, forcing retailers to build custom integrations for every connection. UCP reduces integration complexity and enables structured, machine-readable commerce interactions.

The efficiency gains are obvious. As e-commerce shifts toward agentic experiences, where AI systems handle discovery, comparison, and checkout behind the scenes, standardization becomes table stakes. If a shopper in the Pacific Northwest asks Gemini to find a Seattle Seahawks home jersey in size large of Super Bowl MVP Kenneth Walker III, an AI agent can now scan structured inventory data across multiple retailers and complete the purchase without navigating a maze of bespoke integrations. Of course, AI doesn’t override shopper approval but helps assist in their search.

For merchants, this means continued participation in the transaction even when it doesn’t originate on their owned-and-operated site. Distribution expands. Less friction improves conversations. The path to purchase compresses. But so do traditional trust signals.

The Risk Beneath the Efficiency

UCP is an important step toward making agentic commerce viable at scale. By standardizing how AI agents discover products, make decisions, and initiate checkout, it accelerates automated, delegated shopping. Yet it exposes a structural challenge: identity and intent resolution in agent-initiated transactions.

Today’s fraud prevention systems were built for transactions initiated directly by consumers. They rely on behavioral signals like device characteristics, keyboard language, browsing cadence, merchant network context, and historical identity linkages. These fingerprints help distinguish legitimate buyers from bad actors.

When an AI agent intermediates the purchase, many of those signals diminish or disappear, while identity signals remain foundational. The merchant may not directly observe the device environment. Interaction patterns look different. The “customer” becomes a layer removed from the human shopper.

This doesn't mean merchants are defenseless. But it does require a shift in how trust is established.

From Behavioral Signals to Intent Verification

In agentic commerce, risk assessment can no longer assume the presence of traditional human behavioral signals.

Instead, it must place greater emphasis on validating:

• User intent: Was the transaction explicitly authorized?
• Agent authorization: Is the AI acting within defined permissions with a clearly defined and validated identity?
• Transaction context: Does the purchase align with historical patterns across merchants, categories and prices?
• Cross-merchant intelligence: Are there shared signals indicating coordinated fraud attempts at the network level?

This is a governance challenge as much as a technical one. Merchants will need to define liability boundaries with platforms and agent providers. They'll need to clarify who bears responsibility when intent is ambiguous or compromised. And they'll need to ensure compliance teams understand how AI-mediated transactions alter regulatory exposure.

Building Home-Field Advantage

Forward-thinking brands are likely to deploy their own merchant-side agents — i.e., systems designed to verify intent, validate agent credentials, and secure the customer experience in real time. These agents can strengthen human oversight and layered signals.

Rather than passively accepting an incoming agent transaction, these brands will actively interrogate it by confirming permissions, analyzing cross-channel signals, and reinforcing authentication where necessary.

In effect, they will create a home-field advantage in an increasingly agent-mediated marketplace.

UCP lowers the barrier to interoperability. It makes agentic commerce scalable. However, scalability without trust is fragility at speed. The competitive edge will not simply belong to first-mover retailers. It will belong to those that redesign their trust architecture for a world where machines shop on behalf of humans and where validating intent, which is complementary to identity, is the new frontline of commerce.

Jeff Otto is chief marketing officer at Riskified, a leader in AI-driven fraud prevention and risk intelligence for e-commerce.