As we saw on Black Friday, Amazon’s AI assistant, Rufus is reshaping consumer behavior. Sessions that led to purchases doubled for Rufus users, compared with only a 20% increase for shoppers who didn’t use it.
That shift isn’t siloed to Rufus either. Seventy-three percent of consumers use AI assistants to help them shop, 58% use AI for holiday gifts, and 70% are comfortable letting AI complete their transactions, according to an October Riskified survey of 5,400 people.
The future of AI shopping is likely to follow one of three paths, each with very different implications for retailers:
Merchant-led ecosystem: Some companies will keep AI shopping native to their own platforms, blocking external tools. Amazon’s Rufus is the prototype of this approach. This model preserves control over pricing, customer relationships, and brand experience while allowing fraud teams to maintain the behavioral signals they rely on. The trade-off requires significant investment in AI capabilities to remain competitive.
Collaborative ecosystem: Others will embrace partnerships with third-party AI shopping tools using open protocols such as ACP, AP2, and Visa VIC. Retailers retain some control over pricing and brand presentation while benefiting from AI-driven traffic. These frameworks also allow verification of both consumer and AI interactions, sharing intelligence to reduce fraud and disputes. Google’s AP2 announcement in September 2025, backed by Mastercard, American Express, and PayPal, signals that the industry is moving in this direction. Open standards give retailers a seat at the table and prevent full disintermediation.
Decentralized shopping: The most disruptive path is one in which third-party tools like ChatGPT become the primary interface for consumers. Retailers risk being reduced to undifferentiated inventory suppliers competing on price and fulfillment. Autonomous AI can lead to increased disputes, friendly fraud, and chargebacks, particularly when accounts are compromised. A single hijacked profile can execute fraudulent purchases across multiple stores — a scenario we refer to as “Compromised AI-as-a-Service.”
This agentic evolution is transforming the entire commerce funnel.
Even well-intentioned AI can miss shipping or return policies, generating non-fraud disputes. AI that scans multiple sites for the lowest price can escalate pricing competition. And, with payment security topping consumers’ concerns, retailers can’t afford to be passive.
