Noblis has announced a U.S. patent for an access-control system that continually evaluates biometric and environmental signals, rather than making a single authentication decision when a user signs in. The company says the system is designed to identify impersonation attempts, including photographs and deepfakes, while a session is underway.
The patented approach combines machine learning with observations of the user and the surrounding operating environment. Facial characteristics, gaze and blinking contribute to a liveness score, which the system uses to assess whether the person interacting with a device is present and authentic.
Environmental inputs include spatial-security information, electromagnetic conditions and network integrity. Those signals are evaluated alongside biometrics so that access decisions can respond to a change in the user’s behavior or the security conditions around the device.
Depending on the organization’s policies and the system’s assessment, a user can retain full access, have particular features restricted or be blocked. Noblis describes the technology as a way to adapt electronic access controls during an interaction instead of leaving the original sign-in decision unchanged throughout the session.
The company identifies the award as U.S. Patent 12,682,086. The work was developed through its internal research program, with scientist and inventor Erica Maciejewski among the researchers behind the approach. Noblis says the combination of biometric and contextual evidence is intended to support continued verification in secure environments.
Continuous authentication is also being applied in commercial access products. YEO’s facial-authentication technology has expanded into South Africa through Flownamix, using on-device face and liveness checks to verify that the authorized user remains in a mobile or web application.
A separate ChainIT and RS Software integration for instant payments combines biometric identity tokens with device, behavioral and transaction context. In that application, the payment institution can apply additional controls or hold a transaction based on the combined risk information.
President and chief executive Mile Corrigan said the patent grew out of Noblis’s internal research program, which develops technology before a customer sets a specific requirement. Noblis is a research-driven nonprofit serving federal agencies. Maciejewski’s team trained the machine-learning model to evaluate biometric and environmental inputs continuously and respond when they indicate a possible compromise.
Sources: Noblis
–
By the ID Tech Editorial Team
