Socure has launched a suite designed to identify AI agents, assess their trustworthiness and connect them to the people or businesses authorising them. The key idea is to give organisations more than a bot-or-human verdict when deciding whether to let an agent act.
Watch Desk analysis
What happened
Socure’s Agent Intelligence has three components. Agent Detection, available now within its Digital Intelligence offering, identifies and classifies agents in browser sessions. Agent TrustScore, in beta, separately scores identity, authorisation and risk. Agent Credential, due for general availability in the first quarter of 2027, lets an operator issue a verifiable credential linking an agent to a verified principal.
The company says its detection identified named consumer agents with 99.99% accuracy in controlled testing across 700 behavioural variables. That is Socure’s lab result, not a guarantee of accuracy in every live setting. The company says its trust scoring can also evaluate agents without credentials and credentials from other issuers.
Why it matters
As software acts on people’s behalf, a business needs to know not only whether a session is automated, but which agent is acting, what it is allowed to do and who stands behind it. Socure’s approach aims to let organisations allow, challenge or block activity according to their own risk thresholds, rather than treating every agent as either a welcome customer or a bot to swat.
Our read
This is a concrete attempt to build identity and permission checks for agent-driven activity, rather than merely asking websites to choose between opening the gates and bolting them shut. The useful test will be whether the scores and credentials help organisations make better decisions across different agents, including those Socure does not issue or control.
What to watch
- Whether Agent TrustScore moves from beta into general availability.
- Whether Agent Credential arrives on schedule in Q1 2027.
- How detection and trust scoring perform outside controlled testing, across different agents and real-world sessions.
Discussion spark: Should websites trust an AI agent only when it carries a verifiable link to an accountable person or business, or should other signals be enough?
Sources and evidence
- Source update (27 October 2026, 13:28 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.