iProov releases open HAPS framework verifying genuine human approval behind AI agent actions

September 21, 2026

Global biometric identity verification leader iProov has launched the Human Approval and Presence Specification (HAPS), an open, experimental procedural framework designed to clearly distinguish when a real human has explicitly authorized an action before an AI agent carries it out. Released on GitHub under the Apache-2.0 license, HAPS responds to a critical gap in current AI governance: that an agent having access permissions does not confirm a human actually intended or endorsed its specific actions, whether those stem from prompt injection, overreach, or misused credentials. Unlike high-profile breaches, this quieter risk flies under the radar because receiving systems currently lack reliable ways to confirm request legitimacy. HAPS establishes a standardized way to tie verifiable proof of human approval directly to the exact action being proposed, letting organizations enforce necessary oversight without blocking autonomy entirely. The framework does not demand human sign-off for every routine task—businesses set their own thresholds for which sensitive or high-stakes actions require confirmation. When triggered, the process pauses execution, presents the proposed action clearly to the human, gathers verified evidence of their genuine presence and consent, and validates that evidence before proceeding. Proof methods remain flexible rather than mandated; iProov demonstrates biometric liveness as one robust implementation option, alongside a partial Rust reference version and test vectors to guide adoption. Founder and CEO Andrew Bud noted that as AI shifts from answering questions to acting on our behalf, governance must keep pace, distinguishing permission from confirmed approval and calling on the industry to refine and build on the shared standard together. Bud presented HAPS design principles at AGNTCon + MCPCon Europe in Amsterdam earlier this month. For commerce and enterprise, HAPS delivers tangible value: stronger safeguards reduce operational and reputational risk, create trusted conditions for broader AI adoption across banking, services and regulated sectors, and offer a transparent, collaboratively developed foundation that stakeholders can trust.