Multi-Layer Safety Gate Architecture for Autonomous AI Daemons: Kill-Switch, Confidence Floors, and Weighted Peer Signal Consensus

Read full text Summarise with VIRAT
Authors: Anil Kumar Sharma
Cited by: 0

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

This document is submitted as a defensive publication. No patent rights are asserted. The submission date establishes prior art for each mechanism described. We describe two production-implemented architecture innovations in autonomous AI daemon design, developed in the course of building AnkrOS v3 — a multi-service AI operating system with daemons executing across maritime operations, compliance monitoring, crew management, and financial workflows. The first innovation is the Autonomous Daemon State Machine with Multi-Layer Safety Gates: a five-state finite automaton (SLEEPING → WATCHING → PLANNING → ACTING → NARRATING) with four sequential, non-bypassable safety checkpoints — KarmaTier resolution, ConfidenceFloor enforcement, DharmaOverride category blocking, and EntityLock concurrency control — plus an autonomous kill-switch that fires at a 2% rolling error rate without human intervention, and a loop detector that breaks infinite trigger cascades. The second innovation is the Peer Signal Bus with Confidence Weighting: a Redis Pub/Sub backbone for daemon-to-daemon communication where each signal carries a source_confidence scalar (0.0–1.0), and receiving daemons apply weighted aggregation before deciding to act — weak signals (source_confidence < 0.65) are downweighted by a factor of 0.4, and a coalition of exclusively weak signals is capped below the action threshold regardless of aggregate count. Together, these mechanisms constitute the first publicly documented production system in which autonomous AI agents self-regulate execution autonomy, self-terminate on error thresholds, detect and escalate infinite loops, coordinate via leaderless council synthesis, and apply Byzantine-resistant signal aggregation — all without human intervention at the moment of each decision.

Fields of Study & Subjects

Safety Systems Engineering in Autonomy Distributed systems and fault tolerance Software System Performance and Reliability

Publication details

Journal / source
Zenodo (CERN European Organization for Nuclear Research)
Publication date
25 March 2026
Language
EN
Document type
Preprint