Socure has launched Agent Intelligence, a set of tools that detects AI agents operating in browser sessions, scores whether they should be trusted, and ties each agent to the verified person or business that authorized it.
The product has three components. Agent Detection is generally available within Socure’s existing Digital Intelligence offering, Agent TrustScore is open to beta customers, and Agent Credential is scheduled for general availability in the first quarter of 2027. The tools run on RiskOS, Socure’s orchestration and decisioning platform, which also carries the fraud and compliance automation agents Socure added through its August acquisition of Fravity.
Agent Detection identifies and classifies AI agents driving browser sessions, whether or not they declare themselves, and names the agent when it can be identified. It returns a confidence score and reason codes rather than a yes-or-no verdict, since control of a session can pass between a person and an agent. Socure said its lab testing identified named consumer agents such as Muse and Instinct with 99.99% accuracy across 700 behavioral variables.
Agent TrustScore combines three categories of data: identity, covering who the agent is and who stands behind it; authorization, covering what the agent is permitted to do; and risk, drawn from the histories of the principal and operator, including signs of replay or impersonation. The three results are returned separately. The score evaluates credentials from any issuer, as well as agents that carry no credential.
Agent Credential is a W3C Verifiable Credential, issued through SocurePass, that operators give to their agents to bind each one to a verified principal in Socure’s identity graph. Sites can verify the credential without calling Socure, and the format supports cryptographic chaining and proof of possession so integrity can be carried through a delegation chain when one agent hands work to another.
Socure resolves agents as nodes in the same identity graph that carries SocureID and consumer device identifiers, which the company said handles more than 11 billion identity decisions a year across a customer base that passed 6,000 companies in the third quarter. Organizations can allow, step up or block an agent based on their own risk thresholds.
“Human traffic is going to look like a rounding error next to agent traffic very quickly. But most companies still treat agents like bots and block them,” said Johnny Ayers, CEO of Socure. “The way out isn’t better bot detection. It’s clearly determining if it is an agent, which agent it is, and which human is on the other end of the agent,” he added.
Sources: Help Net Security
–
By the ID Tech Editorial Team
