iProov brings biometric safeguards to autonomous systems

Biometric authentication company iProov has launched a security service designed to prevent autonomous artificial intelligence (AI) agents from carrying out sensitive actions without verified human approval.

Announced on 2 November 2026, the service uses facial verification and cryptographic authentication to confirm that an authorized person has approved a specific action before an AI agent can execute it.

The technology implements the Human Approval and Presence Specification (HAPS), introduced in September 2026, and integrates with open-source AI infrastructure platform agentgateway and authorization management provider Cerbos.

Under the system, an AI agent attempting a high-risk transaction, such as transferring funds, can be required to obtain human authorization.

The designated approver reviews the proposed action and completes a biometric presence check. If approved, the system issues a digitally signed, single-use credential tied to the transaction.

Any changes to the transaction details, including the payment amount or recipient, invalidate the approval.

iProov said the technology provides an auditable record of authorization while helping prevent AI agents from impersonating human decision-makers.

“For organizations to realize the full potential of agentic AI, they need to task agents with increasingly consequential responsibilities. But some actions must remain subject to human control, approval and accountability,” said iProov founder and CEO Andrew Bud.

Cerbos will determine when human approval is required, while iProov verifies the approver’s identity and presence.

The launch comes as businesses increasingly deploy AI agents capable of independently executing tasks, raising concerns about unauthorized transactions, accountability and security.

The HAPS specification is open source, allowing other technology providers to develop compatible approval systems.