What happened
Incognia (announced Oct 6-8) introduced AI Agent Detection for financial institutions: it evaluates an AI agent and the specific action separately — combining signed-request verification, agent-origin classification and bot/automation detection with device/location risk intelligence — plus Web Behavioral Biometrics (mouse, keyboard, copy-paste) and a read-only MCP Server that lets fraud analysts query evidence in natural language.
Why it matters
Represents a new control point for agentic commerce: proving an AI agent is legitimate is distinct from proving a customer authorized a specific action. As Meta/Sierra's Personal Agent Protocol and consumer agents proliferate, this addresses the emerging agent-to-bank verification gap after incidents like the OpenAI-linked agent breach.
Applicability
Financial institutions, marketplaces and mobility firms running agentic payments/account actions; evaluate before enabling agent-initiated transactions at scale.