Industry Pain Points: When AI Shopping Assistants Meet "Black-Box AIs"
As the retail industry fully enters the "Agent Commerce" era, more and more merchants are deploying native conversational AI shopping assistants. However, the other side of the coin is that fraud groups are also using AI agents for automated refund fraud, dealer arbitrage, and promotion abuse.
Riskified, a leader in e-commerce fraud protection, recently announced significant expansions to its AI agent platform. Riskified points out that transaction risks from large model (LLM) traffic are often higher, with risk levels even 2.3 times higher than Google search traffic in certain industries. To address this challenge, Riskified is committed to establishing an authoritative risk intelligence layer between personalized services and security.
Core Upgrades: Dual Protection of Identity Signals and Policy Builder
Assaf Feldman, Chief Technology Officer and co-founder of Riskified, emphasized that the company is acting as a "trust agent," injecting real-time risk intelligence into every critical moment of AI interactions. This upgrade introduced two core features:
AI Agent Identity Signals: This feature uses Riskified's extensive cross-merchant network data to evaluate the real identity behind an AI agent in real time. It helps merchants' AI assistants determine in real time whether a user is eligible for instant refunds or exchanges during conversations, ensuring that every transaction made through AI is trustworthy.
AI Agent Policy Builder: This is a management tool designed specifically for merchants. Merchants can use this tool to configure and implement targeted defense policies, effectively blocking programmatic return abuse and malicious order padding, preventing the AI shopping model from turning from a profit channel into a liability trap.
Strong Collaboration: Building a Trusted Ecosystem for Agent Commerce
In addition to its newly developed features, Riskified also announced a deep collaboration with cybersecurity company HUMAN Security. By combining HUMAN's AgenticTrust technology with Riskified's risk management expertise, merchants can apply unified trust strategies in human and AI-driven interactions.
