Preprint / scholarly article

Hallucination-Aware Audit Gate (HAAG): Observable-Only Action Gating for AI Agents

K. Takahashi

Published
DOI
10.5281/zenodo.18153170

Full text PDF (Zenodo)

Abstract

The preprint defines HAAG, an observable-only audit gate that mediates protected actions via verifiable logs, anchored evidence checkpoints, and capability tokens without relying on model introspection or semantic truth.

Keywords

  • hallucination-aware
  • audit gate
  • observable-only
  • action gating
  • verifiable logs
  • decision anchor
  • capability tokens
  • transparency log
  • AI agents
  • protected actions
  • cryptographic checks

Identifiers and source records