Humanitarian Algorithms : A Codified Key Safety Switch Protocol for Lethal Autonomy

TL;DR

This paper proposes a Humanitarian Algorithms protocol for lethal autonomous weapons, ensuring compliance with International Humanitarian Law through a codified safety switch that incorporates fail-safe mechanisms and machine learning.

Contribution

It introduces a novel safety switch protocol that integrates machine learning to ensure lethal autonomous systems operate within legal and ethical boundaries.

Findings

The protocol enhances safety and legal compliance of autonomous weapons.

It enables fail-safe mechanisms to prevent malfunctioning.

The approach supports the integration of machine learning in autonomous systems.

Abstract

With the deployment of lethal autonomous weapons, there is the requirement that any such platform complies with the precepts of International Humanitarian Law. Humanitarian Algorithms[9: p. 9] ensure that lethal autonomous weapon systems perform military/security operations, within the confines of International Humanitarian Law. Unlike other existing techniques of regulating lethal autonomy this scheme advocates for an approach that enables Machine Learning. Lethal autonomous weapons must be equipped with appropriate fail-safe mechanisms that locks them if they malfunction.