Guidelines for Artificial Intelligence Containment

TL;DR

This paper proposes guidelines for developing AI containment sandboxing software to enhance safety and security when studying and analyzing intelligent artificial agents.

Contribution

It introduces a set of guidelines building on previous work to help AI safety researchers develop reliable containment solutions for various levels of AI.

Findings

Guidelines improve safety in AI sandboxing

Enhance analysis of intelligent agents safely

Reduce risks of information leakage and cyberattacks

Abstract

With almost daily improvements in capabilities of artificial intelligence it is more important than ever to develop safety software for use by the AI research community. Building on our previous work on AI Containment Problem we propose a number of guidelines which should help AI safety researchers to develop reliable sandboxing software for intelligent programs of all levels. Such safety container software will make it possible to study and analyze intelligent artificial agent while maintaining certain level of safety against information leakage, social engineering attacks and cyberattacks from within the container.