FB pixel

Socure launches Agent Intelligence to verify who’s behind AI agents

New suite combines agent detection, identity verification and authorization checks as banks grapple with AI-driven account access
Socure launches Agent Intelligence to verify who’s behind AI agents
 

Socure has launched Agent Intelligence, a suite that detects AI agents, scores whether they should be trusted, and ties each one to the verified person or business that authorized it.

The development comes as major banks acknowledge growing difficulty distinguishing authorized AI agents from other automated activity in customer accounts. Axios reported last week that major U.S. banks are detecting third-party AI agents inside customer accounts, some taking actions that appear independent of user instructions. Many of these agents operate with customer authorization and access credentials, making it difficult for banks to distinguish the agent’s activity from the customer’s own actions. To the bank, they look exactly like the customer. One executive told the outlet the industry struggles to distinguish one agent from another, let alone an agent the customer authorized from one that has gone rogue.

The banks’ response has been blunt. Morgan Stanley has begun blocking third-party agents, while JPMorgan, Bank of America and Citi are restricting them, and most institutions are moving toward read-only access of the kind already granted to budgeting apps. Executives concede they need a middle ground but have not said what it looks like. Socure is betting that the middle ground is an identity problem, and that it already holds most of the answer.

The Agent Intelligence suite runs inside Socure’s RiskOS platform and comes in three layers that the company says go beyond simple Know Your Agent (KYA) tools. Agent Detection is generally available today through Socure’s existing Digital Intelligence product, and current Digital Intelligence customers can switch it on immediately. Agent TrustScore is open to beta customers now. Agent Credential is scheduled to launch in the first quarter of 2027.

A problem growing faster than defenses

The scale behind the launch is striking. Socure says daily requests from AI agents grew by more than 1,700 percent in the year to September 2026. The company also notes that nonhuman traffic now accounts for more than half of internet activity. McKinsey estimates that AI agents could mediate between $3 trillion and $5 trillion of global consumer commerce by 2030.

Consumers appear ready for the shift: Accenture’s 2026 Consumer Pulse report found that 85 percent of consumers were open to collaborating with an AI agent, and 74 percent were open to letting one carry out specific tasks.

Most organizations, Socure argues, are not ready. They cannot see consumer agents in their traffic, do not know how many there are and cannot tell what they are doing. The difficulty compounds because an agent inserts another layer between an action and the person accountable for it, while every verification method in production at scale today was built to evaluate a human.

In Socure’s view, the banks’ current approach of treating this as a human-versus-bot problem and restricting access is a dead end. “Block them all and you block your next customer. Let them all in and you hand fraudsters the keys,” said Johnny Ayers, founder and CEO of Socure. The answer, in his view, is not better bot detection but determining whether a session is an agent, which agent it is and which human stands behind it, which he framed as an evolution of problems Socure has been working on for more than a decade.

Socure contends that existing KYA vendors each tackle only part of the problem. Some detect and classify automated traffic, while others establish an agent’s identity, link it to a principal, or verify credentials and delegated authority. Those capabilities tend to be fragmented, and the company says the vendors lack the identity and fraud intelligence needed to assess both the agent and its principal well enough to make trust decisions. Socure claims Agent Intelligence is the only offering that detects, scores and credentials agents on a single identity graph.

Making the agent a known entity

Socure resolves agents as nodes in the same global identity graph that already carries SocureID and a consumer’s global device identifier, spanning more than 11 billion identity decisions a year. The intent is that an agent stops being an anonymous session and becomes a known entity attached to a known person, with a history.

The first layer, Agent Detection, identifies and classifies AI agents driving browser sessions, whether or not they declare themselves, and names the specific agent when it can. Rather than issuing a binary verdict, it returns a confidence score and reason codes, because control of a session can pass between a human and an agent partway through. Socure says the aim is to go beyond flagging automation and to persistently identify agents, so a relying party can decide whether to trust one rather than block it by default. In controlled lab testing, the company says named consumer agents including Muse and Instinct were identified with 99.99 percent accuracy across roughly 700 behavioral variables.

The second layer, Agent TrustScore, rests on the idea that each sensitive action needs its own trust assessment, even when the agent is tied to a legitimate person or business. The composite score draws on three categories of data. Identity covers who the agent is and who stands behind it. Authorization covers what it is permitted to do and whether the action falls within that permission. Risk covers what the principal’s and operator’s history indicates, including signs of replay or impersonation. The three are returned separately, so customers can weight them according to their use case, and then allow, limit, step up or deny the action under their own policy.

The third layer, Agent Credential, will use W3C Verifiable Credentials issued through Agent Registration. It binds the agent to a KYC- or KYB-verified person or business in Socure’s identity graph and carries evidence of delegated authority. Sites will be able to verify the credential without calling Socure. Because the format supports cryptographic chaining and proof of possession, integrity can be preserved along a delegation chain when one agent hands work to another. The credential is also a pitch to the other side of the market: platforms that deploy agents can issue Socure-backed credentials to make their agents easier to detect, identify and trust, and therefore less likely to be blocked.

Agnostic by design

Socure has deliberately made Agent Intelligence agnostic to agent type, presentation and credential issuer. A trust score that only worked for Socure-issued agents would be useless to a relying party, the company says, so TrustScore evaluates credentials from any issuer as well as agents that carry no credential at all.

Whether the approach holds up will depend on evidence beyond Socure’s own lab tests, on whether agent operators adopt the credential once it ships in 2027, and on how regulators and courts eventually assign liability when an agent the customer trusted does something the customer never asked for.

Related Posts

Article Topics

 |   |   |   | 

Latest Biometrics News

 

Amazon’s facial recognition safeguards face questions over Israeli military surveillance

Amazon Web Services’ (AWS) ability to enforce its own restrictions on biometric surveillance is facing new scrutiny following revelations that…

 

UK begins testing DVS for checks at scale, publishes guidance for data sharing

The UK continues to build out its digital identity ecosystem, with new developments around the Digital Verification Service (DVS) trust…

 

Police deny legal wrongdoing in arrest prompted by false facial recognition match

Authorities in Fargo, North Dakota say they never broke the law in the case of a Tennessee woman who was…

 

New York age assurance certifications signal shift from rules to results

New York State’s SAFE for Kids Act is set to take effect on January 25, 2027, and the Age Check…

 

ICE seeks $1.2 million contract to maintain nationwide license plate surveillance

U.S. Immigration and Customs Enforcement (ICE) is moving to preserve nationwide access to a commercial license plate surveillance system under…

 

Cambodia plans to expand document verification platform into digital ID wallet

Cambodia’s award-winning document verification platform could expand into a much broader digital identity system. Cambodia’s national document verification system, Verify.gov.kh,…

Comments

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Market Intelligence

Biometric Update Podcast

Featured Company

Top Stories This Week

White Papers

Latest Webinars

Biometrics Industry Events