Artificial Intelligence Safety and Cybersecurity: a Timeline of AI Failures

TL;DR

This paper reviews historical AI failures, analyzes their implications for future AI safety, and emphasizes the importance of cybersecurity principles to prevent catastrophic failures in advanced AI systems.

Contribution

It provides a timeline of AI failures, extrapolates future risks, and suggests cybersecurity-inspired safety strategies for AI development.

Findings

AI failures are increasing in frequency and severity.

Safety in narrow AI is comparable to cybersecurity risks.

Superintelligent AI failures could be catastrophic.

Abstract

In this work, we present and analyze reported failures of artificially intelligent systems and extrapolate our analysis to future AIs. We suggest that both the frequency and the seriousness of future AI failures will steadily increase. AI Safety can be improved based on ideas developed by cybersecurity experts. For narrow AIs safety failures are at the same, moderate, level of criticality as in cybersecurity, however for general AI, failures have a fundamentally different impact. A single failure of a superintelligent system may cause a catastrophic event without a chance for recovery. The goal of cybersecurity is to reduce the number of successful attacks on the system; the goal of AI Safety is to make sure zero attacks succeed in bypassing the safety mechanisms. Unfortunately, such a level of performance is unachievable. Every security system will eventually fail; there is no such thing as a 100% secure system.