Socure Launches Agent Intelligence to Verify and Score AI Agents
Socure introduces Agent Intelligence, a new suite designed to detect, score, and link AI agents to their authorized human principals, addressing the growing challenge of distinguishing legitimate AI traffic from malicious activity.
The rapid proliferation of AI agents across the internet has created a significant security challenge, prompting identity verification leader Socure to launch its new Agent Intelligence suite. This platform aims to provide organizations with the tools to detect, score the trustworthiness of, and definitively link AI agents to the verified individuals or businesses that authorized their use. The suite is designed to move beyond basic bot detection and offer a more nuanced understanding of AI-driven interactions.
Agent Intelligence comprises three core components: Agent Detection, Agent TrustScore, and Agent Credential. Agent Detection, now generally available within Socure’s existing Digital Intelligence offering, identifies and classifies AI agents operating within browser sessions, even those that do not self-identify. It provides a confidence score and reason codes, acknowledging that control of a session can fluidly shift between human and AI. In testing, this component achieved 99.99% accuracy in identifying named consumer agents like Muse and Instinct across 700 behavioral variables.
Agent TrustScore, currently in beta, offers a composite risk assessment for AI agents. This score is built from three key data categories: identity (who the agent is and who is behind it), authorization (what the agent is permitted to do and if its actions are within those bounds), and risk (historical data on the principal and operator, including indicators of replay or impersonation). This granular scoring allows organizations to tailor their security responses based on specific use case requirements.
Completing the suite, Agent Credential will reach general availability in Q1 2027. This feature utilizes W3C Verifiable Credentials via SocurePass, enabling operators to issue credentials that cryptographically bind an agent to a verified principal within Socure’s identity graph. These credentials can be verified by any website without direct interaction with Socure, and the format supports cryptographic chaining for delegation scenarios where one agent hands off tasks to another.
The urgency for such a solution is underscored by recent data indicating a massive surge in AI agent activity. In September 2026, daily requests from AI agents increased by over 1,700% year-over-year, with non-human internet traffic surpassing human traffic for the first time. While consumers are increasingly open to collaborating with AI agents for various tasks, most organizations struggle to differentiate this traffic, often mistaking legitimate AI agents for malicious bots and blocking them, or conversely, allowing unchecked access that benefits fraudsters.
Socure’s approach leverages its existing RiskOS platform and global identity graph, which already manages billions of identity decisions annually. By treating AI agents as nodes within this graph, alongside human identities and device identifiers, Socure transforms anonymous sessions into known entities with a traceable history. This allows organizations to implement risk-based controls, such as allowing, stepping up authentication for, or blocking AI agents based on their own defined risk thresholds.
Crucially, Agent Intelligence is designed to be agnostic to the type of agent, its presentation, or the issuer of its credentials. This ensures broad applicability, as a trust score that only functions for Socure-issued agents would be of limited value. The platform's ability to assess credentials from any issuer and agents lacking explicit credentials provides a comprehensive solution for diverse deployment scenarios.
Socure emphasizes that its integrated approach, combining detection, scoring, and credentialing on a single identity graph, offers a distinct advantage over fragmented solutions offered by competitors. The company highlights its market-leading identity graph, advanced AI fraud models, and extensive customer base as key differentiators in addressing the evolving landscape of AI-driven interactions and security threats.