Ant International presented AI SHIELD at a cross-border digital finance compliance conference, a security suite for merchants that combines Anti-Deepfake account protection, the Antom 3-in-1 Transformer payment risk model, a Know-Your-Agent (KYA) identity framework and AgentSafePay, under which Ant says it will fully cover funds lost when a qualifying merchant AI agent makes its own error, such as misreading intent. Ant said its Anti-Deepfake system kept the attack success rate to 2.5% in red-team tests, and that the single Transformer model replaces hundreds of smaller risk models and cuts launch time for new risk scenarios by 80%.
Why it matters: Who pays when an AI agent gets a transaction wrong is one of the biggest open questions in agentic commerce, so a payments provider offering to backstop agent errors is a notable commercial bet. Coordinating KYA with Mastercard and Visa through its Agentic Mobile Protocol also positions Ant International to help set the identity rules for AI agents that pay across networks.
Source: 移动支付网 Mpaypass, October 22, 2026 (Chinese-language source, translated)
Related: Ant International launches Wyn, an AI agent for cross-border merchants on WorldFirst
Image: Siyuwj, CC BY-SA 4.0, via Wikimedia Commons
