iProov Published Human-in-the-Loop AI Protocol
The open-source HAPS framework enables organizations to mandate verifiable human approval before AI agents execute tasks.
Updated on Sept. 18, 2026 in Artificial Intelligence
Live Poll
Should organizations be required to obtain human verification before allowing AI agents to perform critical tasks?
iProov has released the Human Approval and Presence Specification (HAPS), an experimental protocol designed to prevent AI agents from taking unauthorized or unintended actions. The framework is currently available on GitHub as an open-source project.
Why it matters
As AI agents gain increased autonomy, HAPS provides a risk-based standard to distinguish between broad system permissions and specific, verified human authorization. This development aims to formalize how critical operations are gated by human oversight.
The HAPS protocol includes a partial reference implementation written in Rust and standardized test vectors to verify agent compliance. It operates by requiring a programmatic pause in AI execution, necessitating external confirmation before a process can proceed.
The players
iProov
A developer of biometric authentication technology specializing in verifying identity and liveness for digital services.
The details
The HAPS framework functions as a gatekeeping layer that separates AI agent capability from final action execution. By requiring an explicit verification step—which the protocol keeps proof-agnostic, meaning it can interface with various biometric or cryptographic verification methods—it ensures that autonomous agents cannot trigger sensitive commands without a record of human consent. This architecture utilizes a structured request-response cycle that organizations can integrate into their existing agentic workflows to enforce human-in-the-loop oversight.
Timeline
September 18, 2026: iProov officially published the HAPS protocol on GitHub.
The Tech Race
The release of HAPS arrives as organizations move from experimenting with autonomous agents to integrating them into production environments. It competes with various proprietary guardrail systems by offering an open, proof-agnostic standard to ensure human accountability across different AI agent architectures.
Developers and systems architects can now access the HAPS protocol on GitHub to begin testing integration into their existing AI workflows. The framework is licensed under Apache-2.0, allowing organizations to implement and modify the specification to suit their specific security requirements.
The takeaway
iProov is attempting to move AI safety from a conceptual goal to a formalized, technical requirement through its open-source protocol. Future adoption will depend on how widely the industry integrates these specific verification steps into their agentic deployment pipelines.
Further reading
For more on how organizations are managing agentic risk, explore Artificial Intelligence.
Live Poll
Should organizations be required to obtain human verification before allowing AI agents to perform critical tasks?
