US-based identity and risk intelligence provider Socure has introduced Agent Intelligence, a three-layer offering that identifies AI agents, assesses whether they should be trusted, and binds each agent to the verified person or business that authorised it. Agent Detection is generally available from launch within Socure’s existing Digital Intelligence product. Agent TrustScore is open to beta customers, and Agent Credential is scheduled for general availability in Q1 2027.
Agent traffic and the verification gap
The launch reflects the growing volume of automated activity online. Socure cites data showing that by September 2026, daily requests from AI agents had risen by more than 1.700% year on year, and that more than half of internet traffic was non-human for the first time. The company also references Accenture’s 2026 Consumer Pulse report, which found that 85% of consumers were open to collaborating with an AI agent and 74% were open to having an agent carry out specific tasks.
According to Socure, most organisations cannot distinguish traffic generated by consumer agents and have limited visibility into how many agents interact with them or what they do. Agents also add a layer between an action and the person accountable for it, while verification methods currently deployed at scale were designed to assess humans. Johnny Ayers, Founder and CEO of Socure, noted that blocking agents as bots becomes counterproductive once legitimate customers start using consumer agents such as Muse and Instinct. Admitting all agent traffic, by contrast, increases fraud exposure.
Three layers on one identity graph
Agent Detection identifies and classifies AI agents driving browser sessions, whether or not they declare themselves, and names the agent where possible. Instead of a binary verdict, it returns a confidence score and reason codes, as control of a session can pass between a human and an agent. Socure reports that in controlled lab testing, Muse and Instinct were identified with 99.99% accuracy across 700 behavioural variables.
Agent TrustScore is a composite score built from three data categories. These cover identity (the agent and who stands behind it), authorisation (what it is permitted to do), and risk (the histories of the principal and operator, including signs of replay or impersonation). Each category is returned separately so customers can weight attributes according to their use case.
Agent Credential is a W3C Verifiable Credential, issued via SocurePass, that operators assign to their agents to bind them to a verified principal in Socure’s identity graph. Any site can verify the credential without contacting Socure. The format supports cryptographic chaining and proof of possession, allowing integrity to be maintained when one agent delegates work to another.
Positioning in the Know Your Agent market
Socure presents the suite as going beyond existing Know Your Agent (KYA) tools. According to the company, these typically address separate parts of the problem: classifying automated traffic, establishing agent identity, linking agents to principals, or verifying delegated authority.
Agent Intelligence runs within Socure’s RiskOS platform and resolves agents as nodes in the same identity graph that holds SocureID and consumers’ global device identifiers. That graph supports more than 11 billion identity decisions annually. Organisations can allow, step up, or block agent activity based on their own risk thresholds. At the same time, platforms deploying AI agents can issue Socure-backed credentials to make their agents more identifiable and less likely to be blocked.
The suite is agnostic to agent type, presentation, and credential issuer, with Agent TrustScore evaluating credentials from any issuer as well as agents without credentials. Existing Digital Intelligence customers can activate Agent Detection immediately.
