Feedzai expands Digital Trust to detect AI agents at banks

Feedzai has broadened its Digital Trust solution so that banks can identify when AI agents are accessing customer accounts, a capability that lays the groundwork for detecting agentic activity at scale.

The upgrade rests on three pillars: more persistent device recognition, a fraud score informed by Feedzai’s network-wide perspective and pooled intelligence, and cryptographic device binding that establishes a “triangle of trust” linking the customer, their device and the bank.

Each of these elements also stands alone as a practical benefit, allowing banks to strip out needless step-up checks, lower authentication spending and harden defences against multifactor authentication (MFA) bypass.

The release addresses a growing blind spot. A valid login no longer guarantees a trustworthy session, because AI agents can authenticate using genuine credentials, while fraudsters take advantage of compromised devices, account recovery routes and social engineering to manipulate legitimate customers.

Digital Trust draws together endpoint, device, behavioural and network data to evaluate risk both ahead of login and across the full customer journey.

To spot an agent, banks must first understand the device, rely on the signals available and grasp the purpose behind each session.