Riskified, a company specializing in e-commerce fraud and risk intelligence, said Tuesday it is expanding its AI Agent Intelligence platform to help online retailers protect conversational artificial intelligence shopping assistants from fraud and abuse.
The move comes as retailers increase investment in deploying generative AI tools directly on their digital storefronts to enhance customer service and personalize shopping experiences. According to research cited by the company from McKinsey & Company, 82% of retail organizations have launched generative AI pilots focused on reinventing customer service.
Riskified said its expanded platform is designed to serve as a risk intelligence layer for merchants building native AI shopping assistants, helping secure transactions and customer interactions. The company analyzes purchase histories across a global network of ecommerce brands, allowing it to provide data and risk insights individual merchants may not be able to access independently.
“Merchants looking to launch their own virtual shopping assistants have a home-field advantage maintaining direct, personalized relationships with their shoppers,” said Assaf Feldman, chief technology officer and co-founder of Riskified. “Riskified’s role is to serve as the definitive risk intelligence layer to both enhance and secure AI agent interactions. By augmenting a merchant's proprietary customer history with insights from our vast multi-merchant network, we help to ensure that every transaction and claim enacted through a merchant’s AI agent is a verified identity.”
Among the new capabilities is AI Agent Identity Signals, which enables a merchant’s AI shopping assistant to query Riskified’s Identity Graph for risk indicators and resolve customer identities programmatically.
The feature can be integrated through enhancements to the company’s AI Agent Approve offering on AWS Marketplace, via Google’s Agent-to-Agent protocol, or through standard RESTful APIs. Riskified said this allows AI agents to receive real-time risk signals during customer interactions, including when processing refund or exchange requests.
The company also announced enhancements to its AI Agent Policy Builder within its Decision Studio platform. The tool enables merchants to apply business rules to manage order volume and mitigate risks such as refund abuse, reseller arbitrage and promotion misuse originating from conversational AI channels.
Riskified said the expansion reflects growing concerns that fraud rings are targeting early-stage AI protocols and chatbots. By focusing on identity resolution and anomalies in purchasing patterns, the company said it aims to ensure AI-driven shopping experiences remain secure and revenue-generating rather than exposing merchants to new liabilities.